* [PATCH v2 2/5] arm64/mm: drop vmemmap_pmd helpers and use generic code
From: Muchun Song @ 2026-04-04 12:20 UTC (permalink / raw)
To: Catalin Marinas, Will Deacon
Cc: linux-mm, akpm, Muchun Song, Muchun Song, Ryan Roberts,
David Hildenbrand, Kevin Brodsky, Dev Jain, Lorenzo Stoakes,
Anshuman Khandual, Yang Shi, Chaitanya S Prakash,
linux-arm-kernel, linux-kernel
In-Reply-To: <20260404122105.3989557-1-songmuchun@bytedance.com>
The generic implementations now suffice; remove the arm64 copies.
Signed-off-by: Muchun Song <songmuchun@bytedance.com>
---
arch/arm64/mm/mmu.c | 14 --------------
1 file changed, 14 deletions(-)
diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c
index ec1c6971a561..b87053452641 100644
--- a/arch/arm64/mm/mmu.c
+++ b/arch/arm64/mm/mmu.c
@@ -1745,20 +1745,6 @@ static void free_empty_tables(unsigned long addr, unsigned long end,
}
#endif
-void __meminit vmemmap_set_pmd(pmd_t *pmdp, void *p, int node,
- unsigned long addr, unsigned long next)
-{
- pmd_set_huge(pmdp, __pa(p), __pgprot(PROT_SECT_NORMAL));
-}
-
-int __meminit vmemmap_check_pmd(pmd_t *pmdp, int node,
- unsigned long addr, unsigned long next)
-{
- vmemmap_verify((pte_t *)pmdp, node, addr, next);
-
- return pmd_sect(READ_ONCE(*pmdp));
-}
-
int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node,
struct vmem_altmap *altmap)
{
--
2.20.1
^ permalink raw reply related
* [PATCH v2 0/5] mm/sparse-vmemmap: provide generic vmemmap_set_pmd() and vmemmap_check_pmd()
From: Muchun Song @ 2026-04-04 12:20 UTC (permalink / raw)
To: Catalin Marinas, Will Deacon, Huacai Chen, Paul Walmsley,
Palmer Dabbelt, Albert Ou, David S. Miller, Andreas Larsson,
Andrew Morton, David Hildenbrand
Cc: linux-mm, Muchun Song, Muchun Song, WANG Xuerui, Alexandre Ghiti,
Lorenzo Stoakes, Liam R. Howlett, Vlastimil Babka, Mike Rapoport,
Suren Baghdasaryan, Michal Hocko, Ryan Roberts, Kevin Brodsky,
Dev Jain, Anshuman Khandual, Yang Shi, Chaitanya S Prakash,
Petr Tesarik, Vishal Moola (Oracle), Junhui Liu, Austin Kim,
Chengkaitao, Matthew Wilcox (Oracle), Alex Shi, linux-arm-kernel,
linux-kernel, loongarch, linux-riscv, sparclinux
The two weak functions vmemmap_set_pmd() and vmemmap_check_pmd() are
currently no-ops on every architecture, forcing each platform that needs
them to duplicate the same handful of lines. Provide a generic implementation:
- vmemmap_set_pmd() simply sets a huge PMD with PAGE_KERNEL protection.
- vmemmap_check_pmd() verifies that the PMD is present and leaf,
then calls the existing vmemmap_verify() helper.
Architectures that need special handling can continue to override the
weak symbols; everyone else gets the standard version for free.
This series drops the custom implementations in arm64, riscv, loongarch,
and sparc, replacing them with the generic implementation introduced
in the first patch.
v1 -> v2:
- Fixed a tooling issue in v1 where duplicate/conflicting patches
were incorrectly sent to the mailing list. No code changes compared
to the intended v1.
Muchun Song (5):
mm/sparse-vmemmap: provide generic vmemmap_set_pmd() and
vmemmap_check_pmd()
arm64/mm: drop vmemmap_pmd helpers and use generic code
riscv/mm: drop vmemmap_pmd helpers and use generic code
loongarch/mm: drop vmemmap_check_pmd helper and use generic code
sparc/mm: drop vmemmap_check_pmd helper and use generic code
arch/arm64/mm/mmu.c | 14 --------------
arch/loongarch/mm/init.c | 11 -----------
arch/riscv/mm/init.c | 13 -------------
arch/sparc/mm/init_64.c | 11 -----------
mm/sparse-vmemmap.c | 7 ++++++-
5 files changed, 6 insertions(+), 50 deletions(-)
--
2.20.1
^ permalink raw reply
* Re: [PATCH v3 2/2] mailbox: Make mbox_send_message() return error code when tx fails
From: Joonwon Kang @ 2026-04-04 11:47 UTC (permalink / raw)
To: jassisinghbrar
Cc: akpm, angelogioacchino.delregno, jonathanh, joonwonkang,
linux-arm-kernel, linux-kernel, linux-mediatek, linux-tegra,
matthias.bgg, stable, thierry.reding, lee
In-Reply-To: <CABb+yY23aTXeXu6G-8sHjw32DCqmhsJLu2Mt-txenOgTBiyv+A@mail.gmail.com>
> On Fri, Apr 3, 2026 at 10:19 AM Joonwon Kang <joonwonkang@google.com> wrote:
> >
> > > On Thu, Apr 2, 2026 at 12:07 PM Joonwon Kang <joonwonkang@google.com> wrote:
> > > >
> > > > When the mailbox controller failed transmitting message, the error code
> > > > was only passed to the client's tx done handler and not to
> > > > mbox_send_message(). For this reason, the function could return a false
> > > > success. This commit resolves the issue by introducing the tx status and
> > > > checking it before mbox_send_message() returns.
> > > >
> > > Can you please share the scenario when this becomes necessary? This
> > > can potentially change the ground underneath some clients, so we have
> > > to be sure this is really useful.
> >
> > I would say the problem here is generic enough to apply to all the cases where
> > the send result needs to be checked. Since the return value of the send API is
> > not the real send result, any users who believe that this blocking send API
> > will return the real send result could fall for that. For example, users may
> > think the send was successful even though it was not actually. I believe it is
> > uncommon that users have to register a callback solely to get the send result
> > even though they are using the blocking send API already. Also, I guess there
> > is no special reason why only the mailbox send API should work this way among
> > other typical blocking send APIs. For these reasons, this patch makes the send
> > API return the real send result. This way, users will not need to register the
> > redundant callback and I think the return value will align with their common
> > expectation.
> >
> Clients submit a message into the Mailbox subsystem to be sent out to
> the remote side which can happen immediately or later.
> If submission fails, clients get immediately notified. If transmission
> fails (which is now internal to the subsystem) it is reported to the
> client by a callback.
> If the API was called mbox_submit_message (which it actually is)
> instead of mbox_send_message, there would be no confusion.
> We can argue how good/bad the current implementation is, but the fact
> is that it is here. And I am reluctant to cause churn without good
> reason.
> Again, as I said, any, _legal_, setup scenario will help me come over
> my reluctance.
mbox_send_message() in blocking mode is not only for submitting the message in
the first place if we read through the API docs.
From the API doc for `mbox_send_message()`:
```
* If the client had set 'tx_block', the call will return
* either when the remote receives the data or when 'tx_tout' millisecs
* run out.
```
From the API doc for `struct mbox_client`:
```
* @tx_block: If the mbox_send_message should block until data is
* transmitted.
```
With the docs, I think it is apparent that the API covers "transmission" of the
message, not only submission of it. If sumbitting is the sole purpose of the
API, what does the API block for in the first place? We would not need the
blocking mode at all then.
The current return value of the API in failure cases is as follows:
- When submission fails, returns failure.
- When submission succeeds but timeout occurs during transmission, return
failure, i.e. -ETIME.
- When submission succeeds but transmission fails without timeout, return
success.
The third case looks problematic. This patch is focusing on this. There is also
disparity to handle the failure between timeout(the second case) and
non-timeout(the third case). Why does it not return failure when non-timeout
error occurs during transmission whereas it does when timeout occurs during
transmission? If the API is solely for submission, why does it return failure
instead of success in the second case?
In the third case, the controller(or the client) will inform the mailbox core
of the transmission failure. Then, why not return that failure as a return
value of the API despite having this information in the core?
An alternative to fixing this issue would be adding the API doc by saying like:
- In blocking mode, the send API will return -ETIME when timeout occurs during
transmission, but it will not return failure but success(since submission
itself was successful before transmission) when other errors occur during
transmission. Users have to register a callback to collect the error code
when non-timeout error occurs.
But, I think we can go away with this unnecessary confusion by fixing the API
just to return the error code when error occurs regardless of whether it is
timeout or not. Then, we could simply say:
- In blocking mode, the send API will return failure when error occurs.
Since this patch is pointing out this anomaly of the send API's behavior, I
am not sure what scenario we would need more. In other words, the current way
of working would be more surprising to the users than the fixed version of it
as it was when I found out this issue for the first time.
Do you think that this change will cause any other problem on the client side
than fixing the existing issue? If not, I am wondering why not go fix the
issue with this patch.
Thanks,
Joonwon Kang
^ permalink raw reply
* Re: [PATCH 1/2] pmdomain/rockchip: skip QoS operations for idle-only domains
From: Shawn Lin @ 2026-04-04 11:40 UTC (permalink / raw)
To: Daniel Bozeman, finley.xiao, ulf.hansson, heiko, linux-pm,
linux-arm-kernel, linux-rockchip, linux-kernel, Jonas Karlman
Cc: shawn.lin
In-Reply-To: <adAwtiaU-32qjRRE@claude-dev>
+ Jonas
在 2026/04/04 星期六 5:27, Daniel Bozeman 写道:
> I ran both tests you requested:
>
> Test 1: Added pr_err to rockchip_pd_power_on/off to identify
> the crashing domain. With patch 2 only (skip EPROBE_DEFER),
> the crash occurs on PD_VO:
Thanks for fing the PD_VO, and I'm still requesting more docs internally
to check what's going on. I see there are several qos nodes under PD_VO,
but I'm not sure if they all belong to PD_VO and even not sure if their
registers are define correctly.
Perhaps, could you help dig more by removing the qos one by one from
PD_VO to narrow down the broken qos?
I also loop in Jonas who submited the code, to have a look.(I'm also
surprised to see there aren't any Qos nodes under PD_VO in vendor
kernel for reference, but upstream code has...)
>
> rockchip_pd_power_off: vo pwr_mask=0x0
> Internal error: synchronous external abort: 0000000096000010
> Workqueue: pm genpd_power_off_work_fn
> Call trace:
> regmap_mmio_read32le+0x8/0x20
> _regmap_bus_reg_read+0x6c/0xac
> _regmap_read+0x60/0xd8
> regmap_read+0x4c/0x7c
> rockchip_pmu_set_idle_request.isra.0+0x98/0x16c
> rockchip_pd_power+0x130/0x48c
> rockchip_pd_power_off+0x38/0x48
> genpd_power_off.isra.0+0x1f0/0x2f0
> genpd_power_off_work_fn+0x34/0x54
>
> Test 2: Same debug build, booted with clk_ignore_unused
> added to kernel cmdline via U-Boot. Same crash, same domain:
>
> rockchip_pd_power_off: vo pwr_mask=0x0
> Internal error: synchronous external abort: 0000000096000010
> (identical call trace)
>
> The crash occurs even with clk_ignore_unused. The QoS
> registers for PD_VO are inaccessible when genpd attempts
> to power off this idle-only domain.
>
^ permalink raw reply
* Re: [PATCH] ASoC: dt-bindings: rockchip: Convert rk3399-gru-sound to YAML
From: Krzysztof Kozlowski @ 2026-04-04 11:11 UTC (permalink / raw)
To: Anushka Badhe
Cc: Liam Girdwood, Mark Brown, Rob Herring, Krzysztof Kozlowski,
Conor Dooley, Heiko Stuebner, linux-sound, devicetree,
linux-arm-kernel, linux-rockchip, linux-kernel
In-Reply-To: <20260402055635.8798-1-anushkabadhe@gmail.com>
On Thu, Apr 02, 2026 at 11:26:35AM +0530, Anushka Badhe wrote:
> Convert the rockchip,rk3399-gru-sound.txt DT binding to YAML Schema.
DT Schema, not YAML Schema.
Same in subject.
https://elixir.bootlin.com/linux/v6.17-rc3/source/Documentation/devicetree/bindings/submitting-patches.rst#L18
...
> +---
> +$id: http://devicetree.org/schemas/sound/rockchip,rk3399-gru-sound.yaml#
> +$schema: http://devicetree.org/meta-schemas/core.yaml#
> +
> +title: ROCKCHIP with MAX98357A/RT5514/DA7219 codecs on GRU boards
Rockchip
> +
> +maintainers:
> + - Heiko Stuebner <heiko@sntech.de>
> +
> +properties:
> + compatible:
> + const: rockchip,rk3399-gru-sound
> +
> + rockchip,cpu:
> + $ref: /schemas/types.yaml#/definitions/phandle-array
Need to list items. See msm/gpu.yaml,
allwinner,sun4i-a10-display-engine.yaml and others.
And read the driver code to understand what is supposed to be here.
> + description:
> + The phandle of the Rockchip I2S controller that's connected to the codecs
> +
> + rockchip,codec:
> + $ref: /schemas/types.yaml#/definitions/phandle-array
Same here.
> + description: The phandle of the audio codecs
Best regards,
Krzysztof
^ permalink raw reply
* Re: [PATCH v9 0/3] Power Management for Raspberry Pi V3D GPU
From: Maíra Canal @ 2026-04-04 10:52 UTC (permalink / raw)
To: Stefan Wahren, Maxime Ripard, Melissa Wen, Iago Toral Quiroga,
Dave Stevenson, Florian Fainelli, Maíra Canal
Cc: dri-devel, linux-rpi-kernel, linux-arm-kernel,
Broadcom internal kernel review list, kernel-dev, Philipp Zabel
In-Reply-To: <20260331-v3d-power-management-v9-0-f52ff87bfd36@igalia.com>
On Tue, 31 Mar 2026 09:35:50 -0300, Maíra Canal wrote:
> This series introduces Runtime Power Management (PM) support for the
> Raspberry Pi V3D GPU.
>
> Currently, the V3D clock remains enabled for the entire system uptime,
> even when the GPU is idle. With the introduction of Runtime PM, the
> clock can now be disabled during idle periods. For example, with this
> series applied to a Raspberry Pi 5, if we check `vcgencmd measure_clock
> v3d`, we get:
>
> [...]
Applied, thanks for all reviews!
[1/3] drm/v3d: Use devm_reset_control_get_optional_exclusive()
commit: de1e32ef1d625ee4d717bcf10c23df2722324f62
[2/3] drm/v3d: Allocate all resources before enabling the clock
commit: ffd7371ed4179827dcf401543b37b69e5781f924
[3/3] drm/v3d: Introduce Runtime Power Management
commit: 458f2a712ab42b7d3615208862922dc35fe90ef9
Best regards,
- Maíra
^ permalink raw reply
* Re: [PATCH v2] dt-bindings: arm-smmu: qcom: Add compatible for Hawi SoC
From: Krzysztof Kozlowski @ 2026-04-04 10:50 UTC (permalink / raw)
To: Mukesh Ojha
Cc: Will Deacon, Joerg Roedel, Rob Herring, Krzysztof Kozlowski,
Conor Dooley, Robin Murphy, linux-arm-kernel, iommu, devicetree,
linux-kernel
In-Reply-To: <20260403080956.2714415-1-mukesh.ojha@oss.qualcomm.com>
On Fri, Apr 03, 2026 at 01:39:56PM +0530, Mukesh Ojha wrote:
> Qualcomm Hawi SoC include apps smmu that implements arm,mmu-500, which
> is used to translate device-visible virtual addresses to physical
> addresses. Add compatible for these items.
>
> Signed-off-by: Mukesh Ojha <mukesh.ojha@oss.qualcomm.com>
> ---
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com>
Best regards,
Krzysztof
^ permalink raw reply
* Re: (subset) [PATCH v8 0/5] Add i.MX943 PCIe supports
From: Manivannan Sadhasivam @ 2026-04-04 10:42 UTC (permalink / raw)
To: robh, krzk+dt, conor+dt, bhelgaas, frank.li, l.stach, lpieralisi,
kwilczynski, s.hauer, kernel, festevam, Richard Zhu
Cc: linux-pci, linux-arm-kernel, devicetree, imx, linux-kernel
In-Reply-To: <20260324023036.784466-1-hongxing.zhu@nxp.com>
On Tue, 24 Mar 2026 10:30:31 +0800, Richard Zhu wrote:
> This patch-set adds i.MX943 PCIe supports on EVK board. Please pay
> attention to that it relies on the patch-set[1], and the PCIe1 port on
> the EVK board relies on the [2].
>
> Both of them are included in the v7.0 kernel.
> [1] https://lore.kernel.org/imx/176649331066.523506.9443864112044699350.b4-ty@kernel.org/
> [2] https://lore.kernel.org/imx/inzg46tc2fwsajxq4vzdyuiq7krzy6xtcg2mjaieninz7zsmgm@mtdjr4tuegpq/
>
> [...]
Applied, thanks!
[1/5] dt-bindings: PCI: imx6q-pcie: Change maxItems of clocks and clock-names to 6
commit: 401359ef44af43b6b775dc01bb7b31396db67aab
[2/5] dt-bindings: PCI: imx6q-pcie: Add i.MX94 and i.MX943 PCIe compatible strings
commit: 4d7937d8cc32b027a14cb8152d9df64d17e9392c
Best regards,
--
Manivannan Sadhasivam <mani@kernel.org>
^ permalink raw reply
* Re: [PATCH] PCI: aspeed: Fix IRQ domain leak on platform_get_irq() failure
From: Manivannan Sadhasivam @ 2026-04-04 10:35 UTC (permalink / raw)
To: Jacky Chou, Lorenzo Pieralisi, Krzysztof Wilczyński,
Rob Herring, Bjorn Helgaas, Joel Stanley, Andrew Jeffery,
Felix Gu
Cc: linux-aspeed, linux-pci, linux-arm-kernel, linux-kernel
In-Reply-To: <20260324-aspeed-v1-1-354181624c00@gmail.com>
On Tue, 24 Mar 2026 01:57:59 +0800, Felix Gu wrote:
> The aspeed_pcie_probe() function calls aspeed_pcie_init_irq_domain()
> which allocates pcie->intx_domain and initializes MSI. However, if
> platform_get_irq() fails afterwards, the cleanup action was not yet
> registered via devm_add_action_or_reset(), causing the IRQ domain
> resources to leak.
>
> Fix this by registering the devm cleanup action immediately after
> aspeed_pcie_init_irq_domain() succeeds, before calling
> platform_get_irq(). This ensures proper cleanup on any subsequent
> failure.
>
> [...]
Applied, thanks!
[1/1] PCI: aspeed: Fix IRQ domain leak on platform_get_irq() failure
commit: c54d5f5b33990f2649c20f35407f340bcadb8a53
Best regards,
--
Manivannan Sadhasivam <mani@kernel.org>
^ permalink raw reply
* Re: [RFC PATCH 0/2] mm: continue using per-VMA lock when retrying page faults after I/O
From: wang lian @ 2026-04-04 9:19 UTC (permalink / raw)
To: 21cnbao
Cc: akpm, linux-arm-kernel, linux-fsdevel, linux-kernel, linux-mm,
linux-riscv, linux-s390, linuxppc-dev, loongarch, surenb, willy,
wang lian, Wang Lian, Kunwu Chan, Kunwu Chan
In-Reply-To: <CAGsJ_4wnwAet4svDrxT4sTdp24sweAU-2VyYn3iNPOoaKdXxPw@mail.gmail.com>
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=y, Size: 6581 bytes --]
Hi Barry,
> If either you or Matthew have a reproducer for this issue, I’d be
> happy to try it out.
Kunwu and I evaluated this series ("mm: continue using per-VMA lock when
retrying page faults after I/O") under a stress scenario specifically
designed to expose the retry behavior in filemap_fault(). This models
the exact situation described by Matthew Wilcox [1], where retries after
I/O fail to make forward progress under memory pressure.
The scenario targets the critical window between I/O completion and
mmap_lock reacquisition. This workload deliberately includes frequent
mmap/munmap operations to simulate a highly contended mmap_lock
environment alongside severe memory pressure (1GB memcg limit). Under
this pressure, folios instantiated by the I/O can be aggressively
reclaimed before the delayed task can re-acquire the lock and install
the PTE, forcing retries to repeat the entire work.
To make this behavior reproducible, we constructed a stress setup that
intentionally extends this interval:
* 256-core x86 system
* 1GB memory cgroup
* 500 threads continuously faulting on a 16MB file
The core reproducer and the execution command are provided below:
#define _GNU_SOURCE
#include <errno.h>
#include <fcntl.h>
#include <pthread.h>
#include <stdatomic.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/mman.h>
#include <unistd.h>
#include <time.h>
#define THREADS 500
#define FILE_SIZE (16 * 1024 * 1024) /* 16MB */
static _Atomic int g_stop = 0;
#define RUN_SECONDS 600
struct worker_arg {
long id;
uint64_t *counts;
};
void *worker(void *arg)
{
struct worker_arg *wa = (struct worker_arg *)arg;
long id = wa->id;
char path[64];
uint64_t local_rounds = 0;
snprintf(path, sizeof(path), "./test_file_%d_%ld.dat",
getpid(), id);
int fd = open(path, O_RDWR | O_CREAT | O_TRUNC, 0666);
if (fd < 0) return NULL;
if (ftruncate(fd, FILE_SIZE) < 0) {
close(fd); return NULL;
}
while (!atomic_load_explicit(&g_stop, memory_order_relaxed)) {
char *f_map = mmap(NULL, FILE_SIZE, PROT_READ,
MAP_SHARED, fd, 0);
if (f_map != MAP_FAILED) {
/* Pure page cache thrashing */
for (int i = 0; i < FILE_SIZE; i += 4096) {
volatile unsigned char c =
(unsigned char)f_map[i];
(void)c;
}
munmap(f_map, FILE_SIZE);
local_rounds++;
}
}
wa->counts[id] = local_rounds;
close(fd);
unlink(path);
return NULL;
}
int main(void)
{
printf("Pure File Thrashing Started. PID: %d\n", getpid());
pthread_t t[THREADS];
uint64_t local_counts[THREADS];
memset(local_counts, 0, sizeof(local_counts));
struct worker_arg args[THREADS];
for (long i = 0; i < THREADS; i++) {
args[i].id = i;
args[i].counts = local_counts;
pthread_create(&t[i], NULL, worker, &args[i]);
}
sleep(RUN_SECONDS);
atomic_store_explicit(&g_stop, 1, memory_order_relaxed);
for (int i = 0; i < THREADS; i++) pthread_join(t[i], NULL);
uint64_t total = 0;
for (int i = 0; i < THREADS; i++) total += local_counts[i];
printf("Total rounds : %llu\n", (unsigned long long)total);
printf("Throughput : %.2f rounds/sec\n",
(double)total / RUN_SECONDS);
return 0;
}
Command line used for the test:
systemd-run --scope -p MemoryHigh=1G -p MemoryMax=1.2G -p MemorySwapMax=0 \
--unit=mmap-thrash-$$ ./mmap_lock & \
TEST_PID=$!
We also added temporary counters in page fault retries [2]:
- RETRY_IO_MISS : folio not present after I/O completion
- RETRY_MMAP_DROP : retry fallback due to waiting for I/O
We report representative runs from our 600-second test iterations
(kernel v7.0-rc3):
| Case | Total Rounds | Throughput | Miss/Drop(%) | RETRY_MMAP_DROP | RETRY_IO_MISS |
| ------------------- | ------------ | ---------- | ------------ | --------------- | ------------- |
| Baseline (Run 1) | 22,711 | 37.85 /s | 45.04 | 970,078 | 436,956 |
| Baseline (Run 2) | 23,530 | 39.22 /s | 44.96 | 972,043 | 437,077 |
| With Series (Run A) | 54,428 | 90.71 /s | 1.69 | 1,204,124 | 20,398 |
| With Series (Run B) | 35,949 | 59.91 /s | 0.03 | 327,023 | 99 |
Notes:
1. Throughput Improvement: During the 600-second testing window, overall
workload throughput can more than double (e.g., Run A jumped from ~38
to 90.71 rounds/sec).
2. Elimination of Race Condition: Without the patch, ~45% of retries
were invalid because newly fetched folios were evicted during the
mmap_lock reacquisition delay. With the per-VMA retry path, the
invalidation ratio plummeted to near zero (0.03% - 1.69%).
3. Counter Scaling and Variance: In Run A, because the I/O wait
bottleneck is eliminated, the threads advance much faster. Thus, the
absolute number of mmap_lock drops naturally scales up with the
increased throughput. In Run B, the primary bottleneck shifts to the
mmap write-lock contention (lock convoying), causing throughput and
total drops to fluctuate. Crucially, the Miss/Drop ratio remains near
zero regardless of this variance.
Without this series, almost half of the retries fail to observe
completed I/O results, causing severe CPU and I/O waste. With the
finer-grained VMA lock, the faulting threads bypass the heavily
contended mmap_lock entirely during retries, completing the fault
almost instantly.
This scenario perfectly aligns with the exact concern raised, and these
results show that the patch not only successfully eliminates the retry
inefficiency but also tangibly boosts macro-level system throughput.
[1] https://lore.kernel.org/linux-mm/aSip2mWX13sqPW_l@casper.infradead.org/
[2] https://github.com/lianux-mm/ioretry_test/
Tested-by: Wang Lian <wanglian@kylinos.cn>
Tested-by: Kunwu Chan <chentao@kylinos.cn>
Reviewed-by: Wang Lian <lianux.mm@gmail.com>
Reviewed-by: Kunwu Chan <kunwu.chan@gmail.com>
--
Best Regards,
wang lian
^ permalink raw reply
* [PATCH] ARM: atags_compat: bound the deprecated command line copy
From: Pengpeng Hou @ 2026-04-03 8:56 UTC (permalink / raw)
To: linux-arm-kernel; +Cc: linux-kernel, pengpeng
`build_tag_list()` still converts the deprecated `param_struct`
command line with `strlen()` and `strcpy()` from a fixed
`commandline[COMMAND_LINE_SIZE]` array.
That source buffer is not locally proven NUL-terminated before the
conversion runs, so malformed old boot parameters can make the helper
read past the end of the source array while sizing or copying the ATAG
command line.
Use `strnlen()` against the source buffer size and copy the bounded
length with an explicit terminator.
Signed-off-by: Pengpeng Hou <pengpeng@iscas.ac.cn>
---
arch/arm/kernel/atags_compat.c | 7 +++++--
1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/arch/arm/kernel/atags_compat.c b/arch/arm/kernel/atags_compat.c
index 10da11c212cc..aa149710f0c0 100644
--- a/arch/arm/kernel/atags_compat.c
+++ b/arch/arm/kernel/atags_compat.c
@@ -92,6 +92,7 @@ static struct tag * __init memtag(struct tag *tag, unsigned long start, unsigned
static void __init build_tag_list(struct param_struct *params, void *taglist)
{
struct tag *tag = taglist;
+ size_t cmdline_len;
if (params->u1.s.page_size != PAGE_SIZE) {
pr_warn("Warning: bad configuration page, trying to continue\n");
@@ -195,9 +196,11 @@ static void __init build_tag_list(struct param_struct *params, void *taglist)
tag = tag_next(tag);
tag->hdr.tag = ATAG_CMDLINE;
- tag->hdr.size = (strlen(params->commandline) + 3 +
+ cmdline_len = strnlen(params->commandline, sizeof(params->commandline));
+ tag->hdr.size = (cmdline_len + 1 + 3 +
sizeof(struct tag_header)) >> 2;
- strcpy(tag->u.cmdline.cmdline, params->commandline);
+ memcpy(tag->u.cmdline.cmdline, params->commandline, cmdline_len);
+ tag->u.cmdline.cmdline[cmdline_len] = '\0';
tag = tag_next(tag);
tag->hdr.tag = ATAG_NONE;
--
2.50.1 (Apple Git-155)
^ permalink raw reply related
* Re: [PATCH 0/5] mm/sparse-vmemmap: provide generic vmemmap_set_pmd() and vmemmap_check_pmd()
From: Muchun Song @ 2026-04-04 7:35 UTC (permalink / raw)
To: Muchun Song
Cc: Catalin Marinas, Will Deacon, Huacai Chen, Paul Walmsley,
Palmer Dabbelt, Albert Ou, David S. Miller, Andreas Larsson,
Andrew Morton, David Hildenbrand, WANG Xuerui, Alexandre Ghiti,
Lorenzo Stoakes, Liam R. Howlett, Vlastimil Babka, Mike Rapoport,
Suren Baghdasaryan, Michal Hocko, Ryan Roberts, Kevin Brodsky,
Dev Jain, Anshuman Khandual, Yang Shi, Chaitanya S Prakash,
Yuquan Wang, Petr Tesarik, Austin Kim, Vishal Moola (Oracle),
Junhui Liu, Matthew Wilcox (Oracle), Alex Shi, Chengkaitao,
linux-arm-kernel, linux-kernel, loongarch, linux-riscv,
sparclinux, linux-mm
In-Reply-To: <20260404071720.3577290-1-songmuchun@bytedance.com>
> On Apr 4, 2026, at 15:17, Muchun Song <songmuchun@bytedance.com> wrote:
>
> The two weak functions vmemmap_set_pmd() and vmemmap_check_pmd() are
> currently no-ops on every architecture, forcing each platform that needs
> them to duplicate the same handful of lines. Provide a generic implementation:
>
> - vmemmap_set_pmd() simply sets a huge PMD with PAGE_KERNEL protection.
>
> - vmemmap_check_pmd() verifies that the PMD is present and leaf,
> then calls the existing vmemmap_verify() helper.
>
> Architectures that need special handling can continue to override the
> weak symbols; everyone else gets the standard version for free.
>
> This series drops the custom implementations in arm64, riscv, loongarch,
> and sparc, replacing them with the generic implementation introduced
> in the first patch.
>
> Muchun Song (5):
> mm/sparse-vmemmap: provide generic vmemmap_set_pmd() and
> vmemmap_check_pmd()
> arm64/mm: drop vmemmap_pmd helpers and use generic code
> riscv/mm: drop vmemmap_pmd helpers and use generic code
> loongarch/mm: drop vmemmap_check_pmd helper and use generic code
> sparc/mm: drop vmemmap_check_pmd helper and use generic code
Hi all,
Please accept my sincere apologies for the mailing list noise.
Due to an error in my local scripts (failing to clean up the patch
output directory before regenerating the series with an updated commit
range), multiple duplicate and conflicting patches were accidentally
sent to the list simultaneously (10 patches in total instead of the
intended 5).
Sorry again for the inconvenience.
Thanks,
Muchun
>
> arch/arm64/mm/mmu.c | 14 --------------
> arch/loongarch/mm/init.c | 11 -----------
> arch/riscv/mm/init.c | 13 -------------
> arch/sparc/mm/init_64.c | 11 -----------
> mm/sparse-vmemmap.c | 7 ++++++-
> 5 files changed, 6 insertions(+), 50 deletions(-)
>
> --
> 2.20.1
>
^ permalink raw reply
* [PATCH 2/5] arm64/mm: drop vmemmap_pmd helpers and use generic code
From: Muchun Song @ 2026-04-04 7:17 UTC (permalink / raw)
To: Catalin Marinas, Will Deacon
Cc: Muchun Song, Muchun Song, Ryan Roberts, Andrew Morton,
Kevin Brodsky, Dev Jain, Anshuman Khandual, Lorenzo Stoakes,
Yang Shi, Chaitanya S Prakash, linux-arm-kernel, linux-kernel
In-Reply-To: <20260404071720.3577290-1-songmuchun@bytedance.com>
The generic implementations now suffice; remove the arm64 copies.
Signed-off-by: Muchun Song <songmuchun@bytedance.com>
---
arch/arm64/mm/mmu.c | 14 --------------
1 file changed, 14 deletions(-)
diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c
index ec1c6971a561..b87053452641 100644
--- a/arch/arm64/mm/mmu.c
+++ b/arch/arm64/mm/mmu.c
@@ -1745,20 +1745,6 @@ static void free_empty_tables(unsigned long addr, unsigned long end,
}
#endif
-void __meminit vmemmap_set_pmd(pmd_t *pmdp, void *p, int node,
- unsigned long addr, unsigned long next)
-{
- pmd_set_huge(pmdp, __pa(p), __pgprot(PROT_SECT_NORMAL));
-}
-
-int __meminit vmemmap_check_pmd(pmd_t *pmdp, int node,
- unsigned long addr, unsigned long next)
-{
- vmemmap_verify((pte_t *)pmdp, node, addr, next);
-
- return pmd_sect(READ_ONCE(*pmdp));
-}
-
int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node,
struct vmem_altmap *altmap)
{
--
2.20.1
^ permalink raw reply related
* [PATCH 1/4] arm64/mm: drop vmemmap_pmd helpers and use generic code
From: Muchun Song @ 2026-04-04 7:17 UTC (permalink / raw)
To: Catalin Marinas, Will Deacon
Cc: Muchun Song, Muchun Song, Ryan Roberts, Andrew Morton,
David Hildenbrand, Kevin Brodsky, Dev Jain, Lorenzo Stoakes,
Anshuman Khandual, Yang Shi, Chaitanya S Prakash,
linux-arm-kernel, linux-kernel
In-Reply-To: <20260404071720.3577290-1-songmuchun@bytedance.com>
The generic implementations now suffice; remove the arm64 copies.
Signed-off-by: Muchun Song <songmuchun@bytedance.com>
---
arch/arm64/mm/mmu.c | 14 --------------
1 file changed, 14 deletions(-)
diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c
index ec1c6971a561..b87053452641 100644
--- a/arch/arm64/mm/mmu.c
+++ b/arch/arm64/mm/mmu.c
@@ -1745,20 +1745,6 @@ static void free_empty_tables(unsigned long addr, unsigned long end,
}
#endif
-void __meminit vmemmap_set_pmd(pmd_t *pmdp, void *p, int node,
- unsigned long addr, unsigned long next)
-{
- pmd_set_huge(pmdp, __pa(p), __pgprot(PROT_SECT_NORMAL));
-}
-
-int __meminit vmemmap_check_pmd(pmd_t *pmdp, int node,
- unsigned long addr, unsigned long next)
-{
- vmemmap_verify((pte_t *)pmdp, node, addr, next);
-
- return pmd_sect(READ_ONCE(*pmdp));
-}
-
int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node,
struct vmem_altmap *altmap)
{
--
2.20.1
^ permalink raw reply related
* [PATCH 0/5] mm/sparse-vmemmap: provide generic vmemmap_set_pmd() and vmemmap_check_pmd()
From: Muchun Song @ 2026-04-04 7:17 UTC (permalink / raw)
To: Catalin Marinas, Will Deacon, Huacai Chen, Paul Walmsley,
Palmer Dabbelt, Albert Ou, David S. Miller, Andreas Larsson,
Andrew Morton, David Hildenbrand
Cc: Muchun Song, Muchun Song, WANG Xuerui, Alexandre Ghiti,
Lorenzo Stoakes, Liam R. Howlett, Vlastimil Babka, Mike Rapoport,
Suren Baghdasaryan, Michal Hocko, Ryan Roberts, Kevin Brodsky,
Dev Jain, Anshuman Khandual, Yang Shi, Chaitanya S Prakash,
Yuquan Wang, Petr Tesarik, Austin Kim, Vishal Moola (Oracle),
Junhui Liu, Matthew Wilcox (Oracle), Alex Shi, Chengkaitao,
linux-arm-kernel, linux-kernel, loongarch, linux-riscv,
sparclinux, linux-mm
The two weak functions vmemmap_set_pmd() and vmemmap_check_pmd() are
currently no-ops on every architecture, forcing each platform that needs
them to duplicate the same handful of lines. Provide a generic implementation:
- vmemmap_set_pmd() simply sets a huge PMD with PAGE_KERNEL protection.
- vmemmap_check_pmd() verifies that the PMD is present and leaf,
then calls the existing vmemmap_verify() helper.
Architectures that need special handling can continue to override the
weak symbols; everyone else gets the standard version for free.
This series drops the custom implementations in arm64, riscv, loongarch,
and sparc, replacing them with the generic implementation introduced
in the first patch.
Muchun Song (5):
mm/sparse-vmemmap: provide generic vmemmap_set_pmd() and
vmemmap_check_pmd()
arm64/mm: drop vmemmap_pmd helpers and use generic code
riscv/mm: drop vmemmap_pmd helpers and use generic code
loongarch/mm: drop vmemmap_check_pmd helper and use generic code
sparc/mm: drop vmemmap_check_pmd helper and use generic code
arch/arm64/mm/mmu.c | 14 --------------
arch/loongarch/mm/init.c | 11 -----------
arch/riscv/mm/init.c | 13 -------------
arch/sparc/mm/init_64.c | 11 -----------
mm/sparse-vmemmap.c | 7 ++++++-
5 files changed, 6 insertions(+), 50 deletions(-)
--
2.20.1
^ permalink raw reply
* Re: [RFC PATCH v2 3/5] iommu/arm-smmu-v3: Add Stream Table Entry display to debugfs
From: Nicolin Chen @ 2026-04-04 5:43 UTC (permalink / raw)
To: Qinxin Xia
Cc: robin.murphy, will, jpb, linux-arm-kernel, iommu, wangzhou1,
prime.zeng, fanghao11, jonathan.cameron, wuyifan50, linuxarm
In-Reply-To: <20260328101706.3448655-4-xiaqinxin@huawei.com>
On Sat, Mar 28, 2026 at 06:17:04PM +0800, Qinxin Xia wrote:
> +static int smmu_debugfs_ste_show(struct seq_file *seq, void *unused)
> +{
> + struct ste_context *ctx = seq->private;
> + struct arm_smmu_master *master = ctx->master;
> + struct arm_smmu_device *smmu;
> + struct arm_smmu_ste *ste;
> + u32 sid, cfg;
> + int i;
> +
> + if (!master) {
> + seq_puts(seq, "No SMMU master data\n");
> + return 0;
> + }
> +
> + smmu = master->smmu;
> + scoped_guard(mutex, &smmu->streams_mutex) {
Instead:
guard(mutex)(&smmu->streams_mutex);
> + sid = ctx->sid;
> +
> + if (!arm_smmu_sid_in_range(smmu, sid)) {
> + seq_printf(seq, "Invalid Stream ID: %u (max %u)\n",
> + sid, (1 << smmu->sid_bits) - 1);
> + return 0;
> + }
> +
> + ste = arm_smmu_get_step_for_sid(smmu, sid);
> + if (!ste) {
> + seq_printf(seq, "STE not available for SID %u\n", sid);
> + return 0;
> + }
> +
> + seq_printf(seq, "STE for Stream ID %u\n", sid);
> + seq_printf(seq, " Valid: %s\n",
> + le64_to_cpu(ste->data[0]) & STRTAB_STE_0_V ? "Yes" : "No");
> +
> + seq_puts(seq, " Config: ");
> +
> + cfg = FIELD_GET(STRTAB_STE_0_CFG, le64_to_cpu(ste->data[0]));
> +
> + switch (cfg) {
> + case STRTAB_STE_0_CFG_BYPASS:
> + seq_puts(seq, "BYPASS\n");
> + break;
> + case STRTAB_STE_0_CFG_S1_TRANS:
> + seq_puts(seq, "only S1_TRANS\n");
> + break;
> + case STRTAB_STE_0_CFG_S2_TRANS:
> + seq_puts(seq, "only S2_TRANS\n");
> + break;
> + case STRTAB_STE_0_CFG_NESTED:
> + seq_puts(seq, "S1+S2_TRANS\n");
> + break;
> + case STRTAB_STE_0_CFG_ABORT:
> + seq_puts(seq, "ABORT\n");
> + break;
> + default:
> + seq_puts(seq, "UNKNOWN\n");
> + }
> +
> + if (le64_to_cpu(ste->data[0]) & STRTAB_STE_0_CFG_S1_TRANS) {
> + seq_printf(seq, " S1ContextPtr: 0x%016llx\n",
> + le64_to_cpu(ste->data[1]) & STRTAB_STE_0_S1CTXPTR_MASK);
> + }
> +
> + if (le64_to_cpu(ste->data[0]) & STRTAB_STE_0_CFG_S2_TRANS) {
> + seq_printf(seq, " S2ContextPtr: 0x%016llx\n",
> + le64_to_cpu(ste->data[3]) & STRTAB_STE_3_S2TTB_MASK);
> + }
> +
> + /* Display raw STE data */
> + seq_puts(seq, " Raw Data:\n");
> + for (i = 0; i < STRTAB_STE_DWORDS; i++)
> + seq_printf(seq, " STE[%d]: 0x%016llx\n", i,
> + le64_to_cpu(ste->data[i]));
> + }
> + return 0;
Check the indentation of the return line.
> +/**
> + * arm_smmu_debugfs_create_stream_table() - Create debugfs entries for stream table
> + * @dev: device to create entries for
> + * @smmu: SMMU device
> + *
> + * Return: 0 on success, negative error code on failure
> + */
> +int arm_smmu_debugfs_create_stream_table(struct device *dev,
> + struct arm_smmu_device *smmu)
Usually @smmu would be the first parameter.
> +{
> + struct dentry *stream_dir, *dev_dir;
> + struct arm_smmu_master *master;
> + struct ste_context *ctx;
> + char name[64];
> + u32 sid;
> + int i;
> +
> + scoped_guard(mutex, &arm_smmu_debugfs_lock) {
> + if (!smmu->debugfs->stream_dir) {
> + stream_dir = debugfs_create_dir("stream_table",
> + smmu->debugfs->smmu_dir);
> + if (!stream_dir)
> + return -ENOMEM;
> +
> + smmu->debugfs->stream_dir = stream_dir;
> + } else {
> + stream_dir = smmu->debugfs->stream_dir;
> + }
> + }
> +
> + master = dev_iommu_priv_get(dev);
> + if (!master || !master->num_streams)
> + return -ENODEV;
> +
> + for (i = 0; i < master->num_streams; i++) {
> + sid = master->streams[i].id;
> + snprintf(name, sizeof(name), "%u", sid);
> + dev_dir = debugfs_create_dir(name, stream_dir);
> + if (!dev_dir)
> + continue;
> +
> + /* Create STE file */
> + ctx = kzalloc_obj(*ctx);
> + ctx->master = master;
> + ctx->sid = sid;
> + spin_lock(&smmu->debugfs->stream_lock);
> + list_add_tail(&ctx->node, &smmu->debugfs->stream_list);
> + spin_unlock(&smmu->debugfs->stream_lock);
May consider an RCU list instead of locking.
> +/**
> + * arm_smmu_debugfs_remove_stream_table() - Remove debugfs entries for stream table
> + * @dev: device to remove entries for
> + * @smmu: SMMU device
Again, @smmu could be the first parameter.
> + * This function removes the debugfs directories created by
> + * arm_smmu_debugfs_create_stream_table().
> + */
> +void arm_smmu_debugfs_remove_stream_table(struct device *dev,
> + struct arm_smmu_device *smmu)
Please double check the indentation. It looks odd on my side.
> -static struct arm_smmu_ste *
> +#ifndef CONFIG_ARM_SMMU_V3_DEBUGFS
> +static
> +#endif
> +struct arm_smmu_ste *
> arm_smmu_get_step_for_sid(struct arm_smmu_device *smmu, u32 sid)
Could probably move this to the header.
> -static bool arm_smmu_sid_in_range(struct arm_smmu_device *smmu, u32 sid)
> +#ifndef CONFIG_ARM_SMMU_V3_DEBUGFS
> +static
> +#endif
> +bool arm_smmu_sid_in_range(struct arm_smmu_device *smmu, u32 sid)
Ditto
> @@ -3648,10 +3658,14 @@ static void arm_smmu_release_device(struct device *dev)
>
> WARN_ON(master->iopf_refcount);
>
> +#ifdef CONFIG_ARM_SMMU_V3_DEBUGFS
> + arm_smmu_debugfs_remove_stream_table(dev, master->smmu);
> +#endif
> arm_smmu_disable_pasid(master);
> arm_smmu_remove_master(master);
> if (arm_smmu_cdtab_allocated(&master->cd_table))
> arm_smmu_free_cd_tables(master);
> +
> kfree(master);
Meaningless line.
> struct arm_smmu_debugfs {
> + struct list_head stream_list;
> + spinlock_t stream_lock;
> struct dentry *smmu_dir;
> + struct dentry *stream_dir;
> /* Reserved for future extensions */
> };
That's the end of the struct. What do you reserve?
Nicolin
^ permalink raw reply
* [PATCH v3 2/8] perf build loongarch: Remove reference to missing file
From: Ian Rogers @ 2026-04-04 5:40 UTC (permalink / raw)
To: acme
Cc: irogers, 9erthalion6, adrian.hunter, alex, alexander.shishkin,
andrew.jones, aou, atrajeev, blakejones, ctshao, dapeng1.mi,
howardchu95, james.clark, john.g.garry, jolsa, leo.yan,
libunwind-devel, linux-arm-kernel, linux-kernel, linux-perf-users,
linux-riscv, mingo, namhyung, palmer, peterz, pjw, shimin.guo,
tglozar, tmricht, will, yuzhuo
In-Reply-To: <20260404054032.1538095-1-irogers@google.com>
The file was removed in commit e62fae9d9e85 ("perf unwind-libdw: Fix a
cross-arch unwinding bug") but the Build file not updated.
Fixes: e62fae9d9e85 ("perf unwind-libdw: Fix a cross-arch unwinding bug")
Signed-off-by: Ian Rogers <irogers@google.com>
---
tools/perf/arch/loongarch/util/Build | 1 -
1 file changed, 1 deletion(-)
diff --git a/tools/perf/arch/loongarch/util/Build b/tools/perf/arch/loongarch/util/Build
index 3ad73d0289f3..8d91e78d31c9 100644
--- a/tools/perf/arch/loongarch/util/Build
+++ b/tools/perf/arch/loongarch/util/Build
@@ -1,4 +1,3 @@
perf-util-y += header.o
perf-util-$(CONFIG_LOCAL_LIBUNWIND) += unwind-libunwind.o
-perf-util-$(CONFIG_LIBDW_DWARF_UNWIND) += unwind-libdw.o
--
2.53.0.1213.gd9a14994de-goog
^ permalink raw reply related
* [PATCH v3 1/8] perf unwind: Refactor get_entries to allow dynamic libdw/libunwind selection
From: Ian Rogers @ 2026-04-04 5:40 UTC (permalink / raw)
To: acme
Cc: irogers, 9erthalion6, adrian.hunter, alex, alexander.shishkin,
andrew.jones, aou, atrajeev, blakejones, ctshao, dapeng1.mi,
howardchu95, james.clark, john.g.garry, jolsa, leo.yan,
libunwind-devel, linux-arm-kernel, linux-kernel, linux-perf-users,
linux-riscv, mingo, namhyung, palmer, peterz, pjw, shimin.guo,
tglozar, tmricht, will, yuzhuo
In-Reply-To: <20260404054032.1538095-1-irogers@google.com>
Currently, both libdw and libunwind define 'unwind__get_entries'. This
causes a duplicate symbol build failure when both are compiled into
perf.
This commit refactors the DWARF unwind post-processing to be
configurable at runtime via the .perfconfig file option
'unwind.style', or using the argument '--unwind-style' in the commands
'perf report', 'perf script' and 'perf inject', in a similar manner to
the addr2line or the disassembler style.
The file 'tools/perf/util/unwind.c' adds the top-level dispatch
function 'unwind__get_entries'. The backend implementations are
renamed to 'libdw__get_entries' and 'libunwind__get_entries'. Both are
attempted as fallbacks if not configured, or if the primary backend
fails.
Fixes: 2e9191573a69 ("perf build: Remove NO_LIBDW_DWARF_UNWIND option")
Signed-off-by: Ian Rogers <irogers@google.com>
---
tools/perf/builtin-inject.c | 4 ++
tools/perf/builtin-report.c | 4 ++
tools/perf/builtin-script.c | 4 ++
tools/perf/util/Build | 1 +
tools/perf/util/symbol_conf.h | 10 +++
tools/perf/util/unwind-libdw.c | 2 +-
tools/perf/util/unwind-libunwind.c | 2 +-
tools/perf/util/unwind.c | 98 ++++++++++++++++++++++++++++++
tools/perf/util/unwind.h | 63 ++++++++++++-------
9 files changed, 163 insertions(+), 25 deletions(-)
create mode 100644 tools/perf/util/unwind.c
diff --git a/tools/perf/builtin-inject.c b/tools/perf/builtin-inject.c
index 5b29f4296861..9ad681b3c0dc 100644
--- a/tools/perf/builtin-inject.c
+++ b/tools/perf/builtin-inject.c
@@ -26,6 +26,7 @@
#include "util/synthetic-events.h"
#include "util/thread.h"
#include "util/namespaces.h"
+#include "util/unwind.h"
#include "util/util.h"
#include "util/tsc.h"
@@ -2539,6 +2540,9 @@ int cmd_inject(int argc, const char **argv)
OPT_STRING(0, "guestmount", &symbol_conf.guestmount, "directory",
"guest mount directory under which every guest os"
" instance has a subdir"),
+ OPT_CALLBACK(0, "unwind-style", NULL, "unwind style",
+ "unwind styles (libdw,libunwind)",
+ unwind__option),
OPT_BOOLEAN(0, "convert-callchain", &inject.convert_callchain,
"Generate callchains using DWARF and drop register/stack data"),
OPT_END()
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index 95c0bdba6b11..0b0966d94128 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -48,6 +48,7 @@
#include "util/time-utils.h"
#include "util/auxtrace.h"
#include "util/units.h"
+#include "util/unwind.h"
#include "util/util.h" // perf_tip()
#include "ui/ui.h"
#include "ui/progress.h"
@@ -1449,6 +1450,9 @@ int cmd_report(int argc, const char **argv)
OPT_CALLBACK(0, "addr2line-style", NULL, "addr2line style",
"addr2line styles (libdw,llvm,libbfd,addr2line)",
report_parse_addr2line_config),
+ OPT_CALLBACK(0, "unwind-style", NULL, "unwind style",
+ "unwind styles (libdw,libunwind)",
+ unwind__option),
OPT_BOOLEAN(0, "demangle", &symbol_conf.demangle,
"Symbol demangling. Enabled by default, use --no-demangle to disable."),
OPT_BOOLEAN(0, "demangle-kernel", &symbol_conf.demangle_kernel,
diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index 622130d3aed4..ee6131315f96 100644
--- a/tools/perf/builtin-script.c
+++ b/tools/perf/builtin-script.c
@@ -63,6 +63,7 @@
#include <linux/err.h>
#include "util/dlfilter.h"
#include "util/record.h"
+#include "util/unwind.h"
#include "util/util.h"
#include "util/cgroup.h"
#include "util/annotate.h"
@@ -4155,6 +4156,9 @@ int cmd_script(int argc, const char **argv)
"Enable symbol demangling"),
OPT_BOOLEAN(0, "demangle-kernel", &symbol_conf.demangle_kernel,
"Enable kernel symbol demangling"),
+ OPT_CALLBACK(0, "unwind-style", NULL, "unwind style",
+ "unwind styles (libdw,libunwind)",
+ unwind__option),
OPT_STRING(0, "addr2line", &symbol_conf.addr2line_path, "path",
"addr2line binary to use for line numbers"),
OPT_STRING(0, "time", &script.time_str, "str",
diff --git a/tools/perf/util/Build b/tools/perf/util/Build
index 70cc91d00804..01edfccebb88 100644
--- a/tools/perf/util/Build
+++ b/tools/perf/util/Build
@@ -216,6 +216,7 @@ ifndef CONFIG_SETNS
perf-util-y += setns.o
endif
+perf-util-y += unwind.o
perf-util-$(CONFIG_LIBDW) += probe-finder.o
perf-util-$(CONFIG_LIBDW) += dwarf-aux.o
perf-util-$(CONFIG_LIBDW) += dwarf-regs.o
diff --git a/tools/perf/util/symbol_conf.h b/tools/perf/util/symbol_conf.h
index ac1b444a8fd8..004c472cdfb1 100644
--- a/tools/perf/util/symbol_conf.h
+++ b/tools/perf/util/symbol_conf.h
@@ -9,6 +9,15 @@
struct strlist;
struct intlist;
+enum unwind_style {
+ UNWIND_STYLE_UNKNOWN = 0,
+ UNWIND_STYLE_LIBDW,
+ UNWIND_STYLE_LIBUNWIND,
+};
+
+#define MAX_UNWIND_STYLE (UNWIND_STYLE_LIBUNWIND + 1)
+
+
enum a2l_style {
A2L_STYLE_UNKNOWN = 0,
A2L_STYLE_LIBDW,
@@ -80,6 +89,7 @@ struct symbol_conf {
*bt_stop_list_str;
const char *addr2line_path;
enum a2l_style addr2line_style[MAX_A2L_STYLE];
+ enum unwind_style unwind_style[MAX_UNWIND_STYLE];
unsigned long time_quantum;
struct strlist *dso_list,
*comm_list,
diff --git a/tools/perf/util/unwind-libdw.c b/tools/perf/util/unwind-libdw.c
index 05e8e68bd49c..d8a5b7d54192 100644
--- a/tools/perf/util/unwind-libdw.c
+++ b/tools/perf/util/unwind-libdw.c
@@ -339,7 +339,7 @@ frame_callback(Dwfl_Frame *state, void *arg)
DWARF_CB_ABORT : DWARF_CB_OK;
}
-int unwind__get_entries(unwind_entry_cb_t cb, void *arg,
+int libdw__get_entries(unwind_entry_cb_t cb, void *arg,
struct thread *thread,
struct perf_sample *data,
int max_stack,
diff --git a/tools/perf/util/unwind-libunwind.c b/tools/perf/util/unwind-libunwind.c
index cb8be6acfb6f..a0016b897dae 100644
--- a/tools/perf/util/unwind-libunwind.c
+++ b/tools/perf/util/unwind-libunwind.c
@@ -79,7 +79,7 @@ void unwind__finish_access(struct maps *maps)
ops->finish_access(maps);
}
-int unwind__get_entries(unwind_entry_cb_t cb, void *arg,
+int libunwind__get_entries(unwind_entry_cb_t cb, void *arg,
struct thread *thread,
struct perf_sample *data, int max_stack,
bool best_effort)
diff --git a/tools/perf/util/unwind.c b/tools/perf/util/unwind.c
new file mode 100644
index 000000000000..86c2d1692d08
--- /dev/null
+++ b/tools/perf/util/unwind.c
@@ -0,0 +1,98 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "debug.h"
+#include "symbol_conf.h"
+#include "unwind.h"
+#include <linux/string.h>
+#include <string.h>
+#include <stdlib.h>
+
+int unwind__get_entries(unwind_entry_cb_t cb __maybe_unused, void *arg __maybe_unused,
+ struct thread *thread __maybe_unused,
+ struct perf_sample *data __maybe_unused,
+ int max_stack __maybe_unused,
+ bool best_effort __maybe_unused)
+{
+ int ret = 0;
+
+#if defined(HAVE_LIBDW_SUPPORT) || defined(HAVE_LIBUNWIND_SUPPORT)
+ if (symbol_conf.unwind_style[0] == UNWIND_STYLE_UNKNOWN) {
+ int i = 0;
+#ifdef HAVE_LIBDW_SUPPORT
+ symbol_conf.unwind_style[i++] = UNWIND_STYLE_LIBDW;
+#endif
+#ifdef HAVE_LIBUNWIND_SUPPORT
+ symbol_conf.unwind_style[i++] = UNWIND_STYLE_LIBUNWIND;
+#endif
+ }
+#endif //defined(HAVE_LIBDW_SUPPORT) || defined(HAVE_LIBUNWIND_SUPPORT)
+
+ for (size_t i = 0; i < ARRAY_SIZE(symbol_conf.unwind_style); i++) {
+ switch (symbol_conf.unwind_style[i]) {
+ case UNWIND_STYLE_LIBDW:
+ ret = libdw__get_entries(cb, arg, thread, data, max_stack, best_effort);
+ break;
+ case UNWIND_STYLE_LIBUNWIND:
+ ret = libunwind__get_entries(cb, arg, thread, data, max_stack, best_effort);
+ break;
+ case UNWIND_STYLE_UNKNOWN:
+ default:
+#if !defined(HAVE_LIBDW_SUPPORT) && !defined(HAVE_LIBUNWIND_SUPPORT)
+ pr_warning_once(
+ "Error: dwarf unwinding not supported, build perf with libdw or libunwind.\n");
+#endif
+ ret = -1;
+ break;
+ }
+ if (ret == 0)
+ break;
+ }
+ return ret;
+}
+
+int unwind__configure(const char *var, const char *value, void *cb __maybe_unused)
+{
+ static const char * const unwind_style_names[] = {
+ [UNWIND_STYLE_LIBDW] = "libdw",
+ [UNWIND_STYLE_LIBUNWIND] = "libunwind",
+ NULL
+ };
+ char *s, *p, *saveptr;
+ size_t i = 0;
+
+ if (strcmp(var, "unwind.style"))
+ return 0;
+
+ if (!value)
+ return -1;
+
+ s = strdup(value);
+ if (!s)
+ return -1;
+
+ p = strtok_r(s, ",", &saveptr);
+ while (p && i < ARRAY_SIZE(symbol_conf.unwind_style)) {
+ bool found = false;
+ char *q = strim(p);
+
+ for (size_t j = UNWIND_STYLE_LIBDW; j < MAX_UNWIND_STYLE; j++) {
+ if (!strcasecmp(q, unwind_style_names[j])) {
+ symbol_conf.unwind_style[i++] = j;
+ found = true;
+ break;
+ }
+ }
+ if (!found)
+ pr_warning("Unknown unwind style: %s\n", q);
+ p = strtok_r(NULL, ",", &saveptr);
+ }
+
+ free(s);
+ return 0;
+}
+
+int unwind__option(const struct option *opt __maybe_unused,
+ const char *arg,
+ int unset __maybe_unused)
+{
+ return unwind__configure("unwind.style", arg, NULL);
+}
diff --git a/tools/perf/util/unwind.h b/tools/perf/util/unwind.h
index 9f7164c6d9aa..ac0776e39f84 100644
--- a/tools/perf/util/unwind.h
+++ b/tools/perf/util/unwind.h
@@ -4,9 +4,10 @@
#include <linux/compiler.h>
#include <linux/types.h>
-#include "util/map_symbol.h"
+#include "map_symbol.h"
struct maps;
+struct option;
struct perf_sample;
struct thread;
@@ -26,7 +27,9 @@ struct unwind_libunwind_ops {
struct perf_sample *data, int max_stack, bool best_effort);
};
-#ifdef HAVE_DWARF_UNWIND_SUPPORT
+int unwind__configure(const char *var, const char *value, void *cb);
+int unwind__option(const struct option *opt, const char *arg, int unset);
+
/*
* When best_effort is set, don't report errors and fail silently. This could
* be expanded in the future to be more permissive about things other than
@@ -36,8 +39,31 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,
struct thread *thread,
struct perf_sample *data, int max_stack,
bool best_effort);
-/* libunwind specific */
+
+#ifdef HAVE_LIBDW_SUPPORT
+int libdw__get_entries(unwind_entry_cb_t cb, void *arg,
+ struct thread *thread,
+ struct perf_sample *data, int max_stack,
+ bool best_effort);
+#else
+#include "debug.h"
+static inline int libdw__get_entries(unwind_entry_cb_t cb __maybe_unused, void *arg __maybe_unused,
+ struct thread *thread __maybe_unused,
+ struct perf_sample *data __maybe_unused,
+ int max_stack __maybe_unused,
+ bool best_effort __maybe_unused)
+{
+ pr_err("Error: libdw dwarf unwinding not built into perf\n");
+ return -1;
+}
+#endif
+
#ifdef HAVE_LIBUNWIND_SUPPORT
+/* libunwind specific */
+int libunwind__get_entries(unwind_entry_cb_t cb, void *arg,
+ struct thread *thread,
+ struct perf_sample *data, int max_stack,
+ bool best_effort);
#ifndef LIBUNWIND__ARCH_REG_ID
#define LIBUNWIND__ARCH_REG_ID(regnum) libunwind__arch_reg_id(regnum)
#endif
@@ -47,26 +73,16 @@ int unwind__prepare_access(struct maps *maps, struct map *map, bool *initialized
void unwind__flush_access(struct maps *maps);
void unwind__finish_access(struct maps *maps);
#else
-static inline int unwind__prepare_access(struct maps *maps __maybe_unused,
- struct map *map __maybe_unused,
- bool *initialized __maybe_unused)
-{
- return 0;
-}
-
-static inline void unwind__flush_access(struct maps *maps __maybe_unused) {}
-static inline void unwind__finish_access(struct maps *maps __maybe_unused) {}
-#endif
-#else
-static inline int
-unwind__get_entries(unwind_entry_cb_t cb __maybe_unused,
- void *arg __maybe_unused,
- struct thread *thread __maybe_unused,
- struct perf_sample *data __maybe_unused,
- int max_stack __maybe_unused,
- bool best_effort __maybe_unused)
+#include "debug.h"
+static inline int libunwind__get_entries(unwind_entry_cb_t cb __maybe_unused,
+ void *arg __maybe_unused,
+ struct thread *thread __maybe_unused,
+ struct perf_sample *data __maybe_unused,
+ int max_stack __maybe_unused,
+ bool best_effort __maybe_unused)
{
- return 0;
+ pr_err("Error: libunwind dwarf unwinding not built into perf\n");
+ return -1;
}
static inline int unwind__prepare_access(struct maps *maps __maybe_unused,
@@ -78,5 +94,6 @@ static inline int unwind__prepare_access(struct maps *maps __maybe_unused,
static inline void unwind__flush_access(struct maps *maps __maybe_unused) {}
static inline void unwind__finish_access(struct maps *maps __maybe_unused) {}
-#endif /* HAVE_DWARF_UNWIND_SUPPORT */
+#endif
+
#endif /* __UNWIND_H */
--
2.53.0.1213.gd9a14994de-goog
^ permalink raw reply related
* [PATCH v3 8/8] perf unwind-libunwind: Add RISC-V libunwind support
From: Ian Rogers @ 2026-04-04 5:40 UTC (permalink / raw)
To: acme
Cc: irogers, 9erthalion6, adrian.hunter, alex, alexander.shishkin,
andrew.jones, aou, atrajeev, blakejones, ctshao, dapeng1.mi,
howardchu95, james.clark, john.g.garry, jolsa, leo.yan,
libunwind-devel, linux-arm-kernel, linux-kernel, linux-perf-users,
linux-riscv, mingo, namhyung, palmer, peterz, pjw, shimin.guo,
tglozar, tmricht, will, yuzhuo
In-Reply-To: <20260404054032.1538095-1-irogers@google.com>
Add a RISC-V implementation for unwinding.
Signed-off-by: Ian Rogers <irogers@google.com>
---
tools/perf/util/libunwind-arch/Build | 1 +
.../perf/util/libunwind-arch/libunwind-arch.c | 21 ++
.../perf/util/libunwind-arch/libunwind-arch.h | 22 ++
.../util/libunwind-arch/libunwind-riscv.c | 297 ++++++++++++++++++
4 files changed, 341 insertions(+)
create mode 100644 tools/perf/util/libunwind-arch/libunwind-riscv.c
diff --git a/tools/perf/util/libunwind-arch/Build b/tools/perf/util/libunwind-arch/Build
index 87fd657a3248..80d3571918b1 100644
--- a/tools/perf/util/libunwind-arch/Build
+++ b/tools/perf/util/libunwind-arch/Build
@@ -5,6 +5,7 @@ perf-util-$(CONFIG_LIBUNWIND) += libunwind-loongarch.o
perf-util-$(CONFIG_LIBUNWIND) += libunwind-mips.o
perf-util-$(CONFIG_LIBUNWIND) += libunwind-ppc32.o
perf-util-$(CONFIG_LIBUNWIND) += libunwind-ppc64.o
+perf-util-$(CONFIG_LIBUNWIND) += libunwind-riscv.o
perf-util-$(CONFIG_LIBUNWIND) += libunwind-s390.o
perf-util-$(CONFIG_LIBUNWIND) += libunwind-i386.o
perf-util-$(CONFIG_LIBUNWIND) += libunwind-x86_64.o
diff --git a/tools/perf/util/libunwind-arch/libunwind-arch.c b/tools/perf/util/libunwind-arch/libunwind-arch.c
index 8539b4233df4..9a74cf3c8729 100644
--- a/tools/perf/util/libunwind-arch/libunwind-arch.c
+++ b/tools/perf/util/libunwind-arch/libunwind-arch.c
@@ -20,6 +20,8 @@ int get_perf_regnum_for_unw_regnum(unsigned int e_machine, int unw_regnum)
return __get_perf_regnum_for_unw_regnum_ppc32(unw_regnum);
case EM_PPC64:
return __get_perf_regnum_for_unw_regnum_ppc64(unw_regnum);
+ case EM_RISCV:
+ return __get_perf_regnum_for_unw_regnum_riscv(unw_regnum);
case EM_S390:
return __get_perf_regnum_for_unw_regnum_s390(unw_regnum);
case EM_386:
@@ -58,6 +60,9 @@ void libunwind_arch__flush_access(struct maps *maps)
case EM_PPC64:
__libunwind_arch__flush_access_ppc64(maps);
break;
+ case EM_RISCV:
+ __libunwind_arch__flush_access_riscv(maps);
+ break;
case EM_S390:
__libunwind_arch__flush_access_s390(maps);
break;
@@ -98,6 +103,9 @@ void libunwind_arch__finish_access(struct maps *maps)
case EM_PPC64:
__libunwind_arch__finish_access_ppc64(maps);
break;
+ case EM_RISCV:
+ __libunwind_arch__finish_access_riscv(maps);
+ break;
case EM_S390:
__libunwind_arch__finish_access_s390(maps);
break;
@@ -128,6 +136,8 @@ void *libunwind_arch__create_addr_space(unsigned int e_machine)
return __libunwind_arch__create_addr_space_ppc32();
case EM_PPC64:
return __libunwind_arch__create_addr_space_ppc64();
+ case EM_RISCV:
+ return __libunwind_arch__create_addr_space_riscv();
case EM_S390:
return __libunwind_arch__create_addr_space_s390();
case EM_386:
@@ -167,6 +177,9 @@ int libunwind_arch__dwarf_search_unwind_table(unsigned int e_machine,
case EM_PPC64:
return __libunwind_arch__dwarf_search_unwind_table_ppc64(as, ip, di, pi,
need_unwind_info, arg);
+ case EM_RISCV:
+ return __libunwind_arch__dwarf_search_unwind_table_riscv(as, ip, di, pi,
+ need_unwind_info, arg);
case EM_S390:
return __libunwind_arch__dwarf_search_unwind_table_s390(as, ip, di, pi,
need_unwind_info, arg);
@@ -211,6 +224,9 @@ int libunwind_arch__dwarf_find_debug_frame(unsigned int e_machine,
case EM_PPC64:
return __libunwind_arch__dwarf_find_debug_frame_ppc64(found, di_debug, ip, segbase,
obj_name, start, end);
+ case EM_RISCV:
+ return __libunwind_arch__dwarf_find_debug_frame_riscv(found, di_debug, ip, segbase,
+ obj_name, start, end);
case EM_S390:
return __libunwind_arch__dwarf_find_debug_frame_s390(found, di_debug, ip, segbase,
obj_name, start, end);
@@ -250,6 +266,9 @@ struct unwind_info *libunwind_arch_unwind_info__new(struct thread *thread,
case EM_PPC64:
return __libunwind_arch_unwind_info__new_ppc64(thread, sample, max_stack,
best_effort, first_ip);
+ case EM_RISCV:
+ return __libunwind_arch_unwind_info__new_riscv(thread, sample, max_stack,
+ best_effort, first_ip);
case EM_S390:
return __libunwind_arch_unwind_info__new_s390(thread, sample, max_stack,
best_effort, first_ip);
@@ -285,6 +304,8 @@ int libunwind_arch__unwind_step(struct unwind_info *ui)
return __libunwind_arch__unwind_step_ppc32(ui);
case EM_PPC64:
return __libunwind_arch__unwind_step_ppc64(ui);
+ case EM_RISCV:
+ return __libunwind_arch__unwind_step_riscv(ui);
case EM_S390:
return __libunwind_arch__unwind_step_s390(ui);
case EM_386:
diff --git a/tools/perf/util/libunwind-arch/libunwind-arch.h b/tools/perf/util/libunwind-arch/libunwind-arch.h
index 2bf7fc33313b..74a09cd58f38 100644
--- a/tools/perf/util/libunwind-arch/libunwind-arch.h
+++ b/tools/perf/util/libunwind-arch/libunwind-arch.h
@@ -39,6 +39,7 @@ int __get_perf_regnum_for_unw_regnum_loongarch(int unw_regnum);
int __get_perf_regnum_for_unw_regnum_mips(int unw_regnum);
int __get_perf_regnum_for_unw_regnum_ppc32(int unw_regnum);
int __get_perf_regnum_for_unw_regnum_ppc64(int unw_regnum);
+int __get_perf_regnum_for_unw_regnum_riscv(int unw_regnum);
int __get_perf_regnum_for_unw_regnum_s390(int unw_regnum);
int __get_perf_regnum_for_unw_regnum_i386(int unw_regnum);
int __get_perf_regnum_for_unw_regnum_x86_64(int unw_regnum);
@@ -50,6 +51,7 @@ void __libunwind_arch__flush_access_loongarch(struct maps *maps);
void __libunwind_arch__flush_access_mips(struct maps *maps);
void __libunwind_arch__flush_access_ppc32(struct maps *maps);
void __libunwind_arch__flush_access_ppc64(struct maps *maps);
+void __libunwind_arch__flush_access_riscv(struct maps *maps);
void __libunwind_arch__flush_access_s390(struct maps *maps);
void __libunwind_arch__flush_access_i386(struct maps *maps);
void __libunwind_arch__flush_access_x86_64(struct maps *maps);
@@ -61,6 +63,7 @@ void __libunwind_arch__finish_access_loongarch(struct maps *maps);
void __libunwind_arch__finish_access_mips(struct maps *maps);
void __libunwind_arch__finish_access_ppc32(struct maps *maps);
void __libunwind_arch__finish_access_ppc64(struct maps *maps);
+void __libunwind_arch__finish_access_riscv(struct maps *maps);
void __libunwind_arch__finish_access_s390(struct maps *maps);
void __libunwind_arch__finish_access_i386(struct maps *maps);
void __libunwind_arch__finish_access_x86_64(struct maps *maps);
@@ -72,6 +75,7 @@ void *__libunwind_arch__create_addr_space_loongarch(void);
void *__libunwind_arch__create_addr_space_mips(void);
void *__libunwind_arch__create_addr_space_ppc32(void);
void *__libunwind_arch__create_addr_space_ppc64(void);
+void *__libunwind_arch__create_addr_space_riscv(void);
void *__libunwind_arch__create_addr_space_s390(void);
void *__libunwind_arch__create_addr_space_i386(void);
void *__libunwind_arch__create_addr_space_x86_64(void);
@@ -111,6 +115,11 @@ int __libunwind_arch__dwarf_search_unwind_table_ppc64(void *as, uint64_t ip,
void *pi,
int need_unwind_info,
void *arg);
+int __libunwind_arch__dwarf_search_unwind_table_riscv(void *as, uint64_t ip,
+ struct libarch_unwind__dyn_info *di,
+ void *pi,
+ int need_unwind_info,
+ void *arg);
int __libunwind_arch__dwarf_search_unwind_table_s390(void *as, uint64_t ip,
struct libarch_unwind__dyn_info *di,
void *pi,
@@ -176,6 +185,13 @@ int __libunwind_arch__dwarf_find_debug_frame_ppc64(int found,
const char *obj_name,
uint64_t start,
uint64_t end);
+int __libunwind_arch__dwarf_find_debug_frame_riscv(int found,
+ struct libarch_unwind__dyn_info *di_debug,
+ uint64_t ip,
+ uint64_t segbase,
+ const char *obj_name,
+ uint64_t start,
+ uint64_t end);
int __libunwind_arch__dwarf_find_debug_frame_s390(int found,
struct libarch_unwind__dyn_info *di_debug,
uint64_t ip,
@@ -236,6 +252,11 @@ struct unwind_info *__libunwind_arch_unwind_info__new_ppc64(struct thread *threa
int max_stack,
bool best_effort,
uint64_t first_ip);
+struct unwind_info *__libunwind_arch_unwind_info__new_riscv(struct thread *thread,
+ struct perf_sample *sample,
+ int max_stack,
+ bool best_effort,
+ uint64_t first_ip);
struct unwind_info *__libunwind_arch_unwind_info__new_s390(struct thread *thread,
struct perf_sample *sample,
int max_stack,
@@ -266,6 +287,7 @@ int __libunwind_arch__unwind_step_loongarch(struct unwind_info *ui);
int __libunwind_arch__unwind_step_mips(struct unwind_info *ui);
int __libunwind_arch__unwind_step_ppc32(struct unwind_info *ui);
int __libunwind_arch__unwind_step_ppc64(struct unwind_info *ui);
+int __libunwind_arch__unwind_step_riscv(struct unwind_info *ui);
int __libunwind_arch__unwind_step_s390(struct unwind_info *ui);
int __libunwind_arch__unwind_step_i386(struct unwind_info *ui);
int __libunwind_arch__unwind_step_x86_64(struct unwind_info *ui);
diff --git a/tools/perf/util/libunwind-arch/libunwind-riscv.c b/tools/perf/util/libunwind-arch/libunwind-riscv.c
new file mode 100644
index 000000000000..a70a2ea96644
--- /dev/null
+++ b/tools/perf/util/libunwind-arch/libunwind-riscv.c
@@ -0,0 +1,297 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "libunwind-arch.h"
+#include "../debug.h"
+#include "../maps.h"
+#include "../thread.h"
+#include "../../../arch/riscv/include/uapi/asm/perf_regs.h"
+#include <linux/compiler.h>
+#include <linux/kernel.h>
+#include <linux/zalloc.h>
+#include <elf.h>
+#include <errno.h>
+
+#ifdef HAVE_LIBUNWIND_RISCV_SUPPORT
+#include <libunwind-riscv.h>
+#endif
+
+int __get_perf_regnum_for_unw_regnum_riscv(int unw_regnum __maybe_unused)
+{
+#ifndef HAVE_LIBUNWIND_RISCV_SUPPORT
+ return -EINVAL;
+#else
+ switch (unw_regnum) {
+ case UNW_RISCV_X1 ... UNW_RISCV_X31:
+ return unw_regnum - UNW_RISCV_X1 + PERF_REG_RISCV_RA;
+ case UNW_RISCV_PC:
+ return PERF_REG_RISCV_PC;
+ default:
+ pr_err("unwind: invalid reg id %d\n", unw_regnum);
+ return -EINVAL;
+ }
+#endif // HAVE_LIBUNWIND_RISCV_SUPPORT
+}
+
+void __libunwind_arch__flush_access_riscv(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_RISCV_SUPPORT
+ unw_flush_cache(maps__addr_space(maps), 0, 0);
+#endif
+}
+
+void __libunwind_arch__finish_access_riscv(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_RISCV_SUPPORT
+ unw_destroy_addr_space(maps__addr_space(maps));
+#endif
+}
+
+#ifdef HAVE_LIBUNWIND_RISCV_SUPPORT
+static int find_proc_info(unw_addr_space_t as, unw_word_t ip, unw_proc_info_t *pi,
+ int need_unwind_info, void *arg)
+{
+ return __libunwind__find_proc_info(as, ip, pi, need_unwind_info, arg, sizeof(unw_word_t));
+}
+
+static void put_unwind_info(unw_addr_space_t __maybe_unused as,
+ unw_proc_info_t *pi __maybe_unused,
+ void *arg __maybe_unused)
+{
+ pr_debug("unwind: put_unwind_info called\n");
+}
+
+static int get_dyn_info_list_addr(unw_addr_space_t __maybe_unused as,
+ unw_word_t __maybe_unused *dil_addr,
+ void __maybe_unused *arg)
+{
+ return -UNW_ENOINFO;
+}
+
+static int access_mem(unw_addr_space_t as, unw_word_t addr, unw_word_t *valp,
+ int __write, void *arg)
+{
+ return __libunwind__access_mem(as, addr, valp, __write, arg, sizeof(unw_word_t));
+}
+
+static int access_reg(unw_addr_space_t as, unw_regnum_t regnum, unw_word_t *valp,
+ int __write, void *arg)
+{
+ return __libunwind__access_reg(as, regnum, valp, __write, arg, sizeof(unw_word_t));
+}
+
+static int access_fpreg(unw_addr_space_t __maybe_unused as,
+ unw_regnum_t __maybe_unused num,
+ unw_fpreg_t __maybe_unused *val,
+ int __maybe_unused __write,
+ void __maybe_unused *arg)
+{
+ pr_err("unwind: access_fpreg unsupported\n");
+ return -UNW_EINVAL;
+}
+
+static int resume(unw_addr_space_t __maybe_unused as,
+ unw_cursor_t __maybe_unused *cu,
+ void __maybe_unused *arg)
+{
+ pr_err("unwind: resume unsupported\n");
+ return -UNW_EINVAL;
+}
+
+static int get_proc_name(unw_addr_space_t __maybe_unused as,
+ unw_word_t __maybe_unused addr,
+ char __maybe_unused *bufp, size_t __maybe_unused buf_len,
+ unw_word_t __maybe_unused *offp, void __maybe_unused *arg)
+{
+ pr_err("unwind: get_proc_name unsupported\n");
+ return -UNW_EINVAL;
+}
+#endif
+
+void *__libunwind_arch__create_addr_space_riscv(void)
+{
+#ifdef HAVE_LIBUNWIND_RISCV_SUPPORT
+ static unw_accessors_t accessors = {
+ .find_proc_info = find_proc_info,
+ .put_unwind_info = put_unwind_info,
+ .get_dyn_info_list_addr = get_dyn_info_list_addr,
+ .access_mem = access_mem,
+ .access_reg = access_reg,
+ .access_fpreg = access_fpreg,
+ .resume = resume,
+ .get_proc_name = get_proc_name,
+ };
+ unw_addr_space_t addr_space;
+
+ addr_space = unw_create_addr_space(&accessors, /*byte_order=*/0);
+ unw_set_caching_policy(addr_space, UNW_CACHE_GLOBAL);
+ return addr_space;
+#else
+ return NULL;
+#endif
+}
+
+#ifdef HAVE_LIBUNWIND_RISCV_SUPPORT
+extern int UNW_OBJ(dwarf_search_unwind_table) (unw_addr_space_t as,
+ unw_word_t ip,
+ unw_dyn_info_t *di,
+ unw_proc_info_t *pi,
+ int need_unwind_info, void *arg);
+#define dwarf_search_unwind_table UNW_OBJ(dwarf_search_unwind_table)
+#endif
+
+int __libunwind_arch__dwarf_search_unwind_table_riscv(void *as __maybe_unused,
+ uint64_t ip __maybe_unused,
+ struct libarch_unwind__dyn_info *_di __maybe_unused,
+ void *pi __maybe_unused,
+ int need_unwind_info __maybe_unused,
+ void *arg __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_RISCV_SUPPORT
+ unw_dyn_info_t di = {
+ .format = UNW_INFO_FORMAT_REMOTE_TABLE,
+ .start_ip = _di->start_ip,
+ .end_ip = _di->end_ip,
+ .u = {
+ .rti = {
+ .segbase = _di->segbase,
+ .table_data = _di->table_data,
+ .table_len = _di->table_len,
+ },
+ },
+ };
+ int ret = dwarf_search_unwind_table(as, ip, &di, pi, need_unwind_info, arg);
+
+ _di->start_ip = di.start_ip;
+ _di->end_ip = di.end_ip;
+ _di->segbase = di.u.rti.segbase;
+ _di->table_data = di.u.rti.table_data;
+ _di->table_len = di.u.rti.table_len;
+ return ret;
+#else
+ return -EINVAL;
+#endif
+}
+
+#if defined(HAVE_LIBUNWIND_RISCV_SUPPORT) && !defined(NO_LIBUNWIND_DEBUG_FRAME_RISCV)
+extern int UNW_OBJ(dwarf_find_debug_frame) (int found, unw_dyn_info_t *di_debug,
+ unw_word_t ip,
+ unw_word_t segbase,
+ const char *obj_name, unw_word_t start,
+ unw_word_t end);
+#define dwarf_find_debug_frame UNW_OBJ(dwarf_find_debug_frame)
+#endif
+
+int __libunwind_arch__dwarf_find_debug_frame_riscv(int found __maybe_unused,
+ struct libarch_unwind__dyn_info *_di __maybe_unused,
+ uint64_t ip __maybe_unused,
+ uint64_t segbase __maybe_unused,
+ const char *obj_name __maybe_unused,
+ uint64_t start __maybe_unused,
+ uint64_t end __maybe_unused)
+{
+#if defined(HAVE_LIBUNWIND_RISCV_SUPPORT) && !defined(NO_LIBUNWIND_DEBUG_FRAME_RISCV)
+ unw_dyn_info_t di = {
+ .format = UNW_INFO_FORMAT_REMOTE_TABLE,
+ .start_ip = _di->start_ip,
+ .end_ip = _di->end_ip,
+ .u = {
+ .rti = {
+ .segbase = _di->segbase,
+ .table_data = _di->table_data,
+ .table_len = _di->table_len,
+ },
+ },
+ };
+ int ret = dwarf_find_debug_frame(found, &di, ip, segbase, obj_name, start, end);
+
+ _di->start_ip = di.start_ip;
+ _di->end_ip = di.end_ip;
+ _di->segbase = di.u.rti.segbase;
+ _di->table_data = di.u.rti.table_data;
+ _di->table_len = di.u.rti.table_len;
+ return ret;
+#else
+ return -EINVAL;
+#endif
+}
+
+struct unwind_info *__libunwind_arch_unwind_info__new_riscv(struct thread *thread __maybe_unused,
+ struct perf_sample *sample __maybe_unused,
+ int max_stack __maybe_unused,
+ bool best_effort __maybe_unused,
+ uint64_t first_ip __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_RISCV_SUPPORT
+ struct arch_unwind_info {
+ struct unwind_info ui;
+ unw_cursor_t _cursor;
+ uint64_t _ips[];
+ };
+
+ struct maps *maps = thread__maps(thread);
+ void *addr_space = maps__addr_space(maps);
+ struct arch_unwind_info *ui;
+ int ret;
+
+ if (addr_space == NULL)
+ return NULL;
+
+ ui = zalloc(sizeof(*ui) + sizeof(ui->_ips[0]) * max_stack);
+ if (!ui)
+ return NULL;
+
+ ui->ui.machine = maps__machine(maps);
+ ui->ui.thread = thread;
+ ui->ui.sample = sample;
+ ui->ui.cursor = &ui->_cursor;
+ ui->ui.ips = &ui->_ips[0];
+ ui->ui.ips[0] = first_ip;
+ ui->ui.cur_ip = 1;
+ ui->ui.max_ips = max_stack;
+ ui->ui.unw_word_t_size = sizeof(unw_word_t);
+ ui->ui.e_machine = EM_RISCV;
+ ui->ui.best_effort = best_effort;
+
+ ret = unw_init_remote(&ui->_cursor, addr_space, &ui->ui);
+ if (ret) {
+ if (!best_effort)
+ pr_err("libunwind: %s\n", unw_strerror(ret));
+ free(ui);
+ return NULL;
+ }
+
+ return &ui->ui;
+#else
+ return NULL;
+#endif
+}
+
+int __libunwind_arch__unwind_step_riscv(struct unwind_info *ui __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_RISCV_SUPPORT
+ int ret;
+
+ if (ui->cur_ip >= ui->max_ips)
+ return -1;
+
+ ret = unw_step(ui->cursor);
+ if (ret > 0) {
+ uint64_t ip;
+
+ unw_get_reg(ui->cursor, UNW_REG_IP, &ip);
+
+ if (unw_is_signal_frame(ui->cursor) <= 0) {
+ /*
+ * Decrement the IP for any non-activation frames. This
+ * is required to properly find the srcline for caller
+ * frames. See also the documentation for
+ * dwfl_frame_pc(), which this code tries to replicate.
+ */
+ --ip;
+ }
+ ui->ips[ui->cur_ip++] = ip;
+ }
+ return ret;
+#else
+ return -EINVAL;
+#endif
+}
--
2.53.0.1213.gd9a14994de-goog
^ permalink raw reply related
* [PATCH v3 6/8] perf unwind-libunwind: Move flush/finish access out of local
From: Ian Rogers @ 2026-04-04 5:40 UTC (permalink / raw)
To: acme
Cc: irogers, 9erthalion6, adrian.hunter, alex, alexander.shishkin,
andrew.jones, aou, atrajeev, blakejones, ctshao, dapeng1.mi,
howardchu95, james.clark, john.g.garry, jolsa, leo.yan,
libunwind-devel, linux-arm-kernel, linux-kernel, linux-perf-users,
linux-riscv, mingo, namhyung, palmer, peterz, pjw, shimin.guo,
tglozar, tmricht, will, yuzhuo
In-Reply-To: <20260404054032.1538095-1-irogers@google.com>
Flush and finish access are relatively simple calls into libunwind,
move them out struct unwind_libunwind_ops. So that the correct version
can be called, add an e_machine variable to maps. This size regression
will go away when the unwind_libunwind_ops no longer need stashing in
the maps. To set the e_machine up pass it into unwind__prepare_access,
which no longer needs to determine the unwind operations based on a
map dso because of this. This also means the maps copying code can
call unwind__prepare_access once for the e_machine rather than once
per map.
Signed-off-by: Ian Rogers <irogers@google.com>
---
.../perf/util/libunwind-arch/libunwind-arch.c | 82 +++++++++++++++++++
.../perf/util/libunwind-arch/libunwind-arch.h | 24 ++++++
.../perf/util/libunwind-arch/libunwind-arm.c | 19 +++++
.../util/libunwind-arch/libunwind-arm64.c | 20 +++++
.../perf/util/libunwind-arch/libunwind-i386.c | 15 ++++
.../util/libunwind-arch/libunwind-loongarch.c | 15 ++++
.../perf/util/libunwind-arch/libunwind-mips.c | 15 ++++
.../util/libunwind-arch/libunwind-ppc32.c | 15 ++++
.../util/libunwind-arch/libunwind-ppc64.c | 15 ++++
.../perf/util/libunwind-arch/libunwind-s390.c | 15 ++++
.../util/libunwind-arch/libunwind-x86_64.c | 15 ++++
tools/perf/util/maps.c | 31 ++++---
tools/perf/util/maps.h | 2 +
tools/perf/util/thread.c | 29 +------
tools/perf/util/unwind-libunwind-local.c | 12 ---
tools/perf/util/unwind-libunwind.c | 61 +++++---------
tools/perf/util/unwind.h | 8 +-
17 files changed, 299 insertions(+), 94 deletions(-)
diff --git a/tools/perf/util/libunwind-arch/libunwind-arch.c b/tools/perf/util/libunwind-arch/libunwind-arch.c
index 5439bf90d161..9692e6c81492 100644
--- a/tools/perf/util/libunwind-arch/libunwind-arch.c
+++ b/tools/perf/util/libunwind-arch/libunwind-arch.c
@@ -1,6 +1,7 @@
// SPDX-License-Identifier: GPL-2.0
#include "libunwind-arch.h"
#include "../debug.h"
+#include "../maps.h"
#include <elf.h>
#include <errno.h>
@@ -30,3 +31,84 @@ int get_perf_regnum_for_unw_regnum(unsigned int e_machine, int unw_regnum)
return -EINVAL;
}
}
+
+
+void libunwind_arch__flush_access(struct maps *maps)
+{
+ unsigned int e_machine = maps__e_machine(maps);
+
+ switch (e_machine) {
+ case EM_NONE:
+ break; // No libunwind info on the maps.
+ case EM_ARM:
+ __libunwind_arch__flush_access_arm(maps);
+ break;
+ case EM_AARCH64:
+ __libunwind_arch__flush_access_arm64(maps);
+ break;
+ case EM_LOONGARCH:
+ __libunwind_arch__flush_access_loongarch(maps);
+ break;
+ case EM_MIPS:
+ __libunwind_arch__flush_access_mips(maps);
+ break;
+ case EM_PPC:
+ __libunwind_arch__flush_access_ppc32(maps);
+ break;
+ case EM_PPC64:
+ __libunwind_arch__flush_access_ppc64(maps);
+ break;
+ case EM_S390:
+ __libunwind_arch__flush_access_s390(maps);
+ break;
+ case EM_386:
+ __libunwind_arch__flush_access_i386(maps);
+ break;
+ case EM_X86_64:
+ __libunwind_arch__flush_access_x86_64(maps);
+ break;
+ default:
+ pr_err("ELF MACHINE %x is not supported.\n", e_machine);
+ break;
+ }
+}
+
+void libunwind_arch__finish_access(struct maps *maps)
+{
+ unsigned int e_machine = maps__e_machine(maps);
+
+ switch (e_machine) {
+ case EM_NONE:
+ break; // No libunwind info on the maps.
+ case EM_ARM:
+ __libunwind_arch__finish_access_arm(maps);
+ break;
+ case EM_AARCH64:
+ __libunwind_arch__finish_access_arm64(maps);
+ break;
+ case EM_LOONGARCH:
+ __libunwind_arch__finish_access_loongarch(maps);
+ break;
+ case EM_MIPS:
+ __libunwind_arch__finish_access_mips(maps);
+ break;
+ case EM_PPC:
+ __libunwind_arch__finish_access_ppc32(maps);
+ break;
+ case EM_PPC64:
+ __libunwind_arch__finish_access_ppc64(maps);
+ break;
+ case EM_S390:
+ __libunwind_arch__finish_access_s390(maps);
+ break;
+ case EM_386:
+ __libunwind_arch__finish_access_i386(maps);
+ break;
+ case EM_X86_64:
+ __libunwind_arch__finish_access_x86_64(maps);
+ break;
+ default:
+ pr_err("ELF MACHINE %x is not supported.\n", e_machine);
+ break;
+ }
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-arch.h b/tools/perf/util/libunwind-arch/libunwind-arch.h
index e1009c6cb965..c00277a5e914 100644
--- a/tools/perf/util/libunwind-arch/libunwind-arch.h
+++ b/tools/perf/util/libunwind-arch/libunwind-arch.h
@@ -2,6 +2,8 @@
#ifndef __LIBUNWIND_ARCH_H
#define __LIBUNWIND_ARCH_H
+struct maps;
+
int __get_perf_regnum_for_unw_regnum_arm(int unw_regnum);
int __get_perf_regnum_for_unw_regnum_arm64(int unw_regnum);
int __get_perf_regnum_for_unw_regnum_loongarch(int unw_regnum);
@@ -13,4 +15,26 @@ int __get_perf_regnum_for_unw_regnum_i386(int unw_regnum);
int __get_perf_regnum_for_unw_regnum_x86_64(int unw_regnum);
int get_perf_regnum_for_unw_regnum(unsigned int e_machine, int unw_regnum);
+void __libunwind_arch__flush_access_arm(struct maps *maps);
+void __libunwind_arch__flush_access_arm64(struct maps *maps);
+void __libunwind_arch__flush_access_loongarch(struct maps *maps);
+void __libunwind_arch__flush_access_mips(struct maps *maps);
+void __libunwind_arch__flush_access_ppc32(struct maps *maps);
+void __libunwind_arch__flush_access_ppc64(struct maps *maps);
+void __libunwind_arch__flush_access_s390(struct maps *maps);
+void __libunwind_arch__flush_access_i386(struct maps *maps);
+void __libunwind_arch__flush_access_x86_64(struct maps *maps);
+void libunwind_arch__flush_access(struct maps *maps);
+
+void __libunwind_arch__finish_access_arm(struct maps *maps);
+void __libunwind_arch__finish_access_arm64(struct maps *maps);
+void __libunwind_arch__finish_access_loongarch(struct maps *maps);
+void __libunwind_arch__finish_access_mips(struct maps *maps);
+void __libunwind_arch__finish_access_ppc32(struct maps *maps);
+void __libunwind_arch__finish_access_ppc64(struct maps *maps);
+void __libunwind_arch__finish_access_s390(struct maps *maps);
+void __libunwind_arch__finish_access_i386(struct maps *maps);
+void __libunwind_arch__finish_access_x86_64(struct maps *maps);
+void libunwind_arch__finish_access(struct maps *maps);
+
#endif /* __LIBUNWIND_ARCH_H */
diff --git a/tools/perf/util/libunwind-arch/libunwind-arm.c b/tools/perf/util/libunwind-arch/libunwind-arm.c
index 6740ee55b043..bbaf01406c52 100644
--- a/tools/perf/util/libunwind-arch/libunwind-arm.c
+++ b/tools/perf/util/libunwind-arch/libunwind-arm.c
@@ -1,10 +1,15 @@
// SPDX-License-Identifier: GPL-2.0
#include "libunwind-arch.h"
#include "../debug.h"
+#include "../maps.h"
#include "../../../arch/arm/include/uapi/asm/perf_regs.h"
#include <linux/compiler.h>
#include <errno.h>
+#ifdef HAVE_LIBUNWIND_ARM_SUPPORT
+#include <libunwind-arm.h>
+#endif
+
int __get_perf_regnum_for_unw_regnum_arm(int unw_regnum)
{
if (unw_regnum < 0 || unw_regnum >= PERF_REG_ARM_MAX) {
@@ -13,3 +18,17 @@ int __get_perf_regnum_for_unw_regnum_arm(int unw_regnum)
}
return unw_regnum;
}
+
+void __libunwind_arch__flush_access_arm(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_ARM_SUPPORT
+ unw_flush_cache(maps__addr_space(maps), 0, 0);
+#endif
+}
+
+void __libunwind_arch__finish_access_arm(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_ARM_SUPPORT
+ unw_destroy_addr_space(maps__addr_space(maps));
+#endif
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-arm64.c b/tools/perf/util/libunwind-arch/libunwind-arm64.c
index 53b1877dfa04..8ba510089736 100644
--- a/tools/perf/util/libunwind-arch/libunwind-arm64.c
+++ b/tools/perf/util/libunwind-arch/libunwind-arm64.c
@@ -1,9 +1,15 @@
// SPDX-License-Identifier: GPL-2.0
#include "libunwind-arch.h"
#include "../debug.h"
+#include "../maps.h"
#include "../../../arch/arm64/include/uapi/asm/perf_regs.h"
+#include <linux/compiler.h>
#include <errno.h>
+#ifdef HAVE_LIBUNWIND_AARCH64_SUPPORT
+#include <libunwind-aarch64.h>
+#endif
+
int __get_perf_regnum_for_unw_regnum_arm64(int unw_regnum)
{
if (unw_regnum < 0 || unw_regnum >= PERF_REG_ARM64_EXTENDED_MAX) {
@@ -12,3 +18,17 @@ int __get_perf_regnum_for_unw_regnum_arm64(int unw_regnum)
}
return unw_regnum;
}
+
+void __libunwind_arch__flush_access_arm64(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_AARCH64_SUPPORT
+ unw_flush_cache(maps__addr_space(maps), 0, 0);
+#endif
+}
+
+void __libunwind_arch__finish_access_arm64(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_AARCH64_SUPPORT
+ unw_destroy_addr_space(maps__addr_space(maps));
+#endif
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-i386.c b/tools/perf/util/libunwind-arch/libunwind-i386.c
index 9eaf4ebff0c2..45ff30c95c1b 100644
--- a/tools/perf/util/libunwind-arch/libunwind-i386.c
+++ b/tools/perf/util/libunwind-arch/libunwind-i386.c
@@ -1,6 +1,7 @@
// SPDX-License-Identifier: GPL-2.0
#include "libunwind-arch.h"
#include "../debug.h"
+#include "../maps.h"
#include "../../../arch/x86/include/uapi/asm/perf_regs.h"
#include <linux/compiler.h>
#include <linux/kernel.h>
@@ -41,3 +42,17 @@ int __get_perf_regnum_for_unw_regnum_i386(int unw_regnum __maybe_unused)
return perf_i386_regnums[unw_regnum];
#endif // HAVE_LIBUNWIND_X86_SUPPORT
}
+
+void __libunwind_arch__flush_access_i386(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_X86_SUPPORT
+ unw_flush_cache(maps__addr_space(maps), 0, 0);
+#endif
+}
+
+void __libunwind_arch__finish_access_i386(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_X86_SUPPORT
+ unw_destroy_addr_space(maps__addr_space(maps));
+#endif
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-loongarch.c b/tools/perf/util/libunwind-arch/libunwind-loongarch.c
index 7009410989bc..837aa11e2b9f 100644
--- a/tools/perf/util/libunwind-arch/libunwind-loongarch.c
+++ b/tools/perf/util/libunwind-arch/libunwind-loongarch.c
@@ -1,6 +1,7 @@
// SPDX-License-Identifier: GPL-2.0
#include "libunwind-arch.h"
#include "../debug.h"
+#include "../maps.h"
#include "../../../arch/loongarch/include/uapi/asm/perf_regs.h"
#include <linux/compiler.h>
#include <errno.h>
@@ -25,3 +26,17 @@ int __get_perf_regnum_for_unw_regnum_loongarch(int unw_regnum __maybe_unused)
}
#endif // HAVE_LIBUNWIND_LOONGARCH64_SUPPORT
}
+
+void __libunwind_arch__flush_access_loongarch(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_LOONGARCH64_SUPPORT
+ unw_flush_cache(maps__addr_space(maps), 0, 0);
+#endif
+}
+
+void __libunwind_arch__finish_access_loongarch(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_LOONGARCH64_SUPPORT
+ unw_destroy_addr_space(maps__addr_space(maps));
+#endif
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-mips.c b/tools/perf/util/libunwind-arch/libunwind-mips.c
index 01a506c8079c..1fa81742ff4a 100644
--- a/tools/perf/util/libunwind-arch/libunwind-mips.c
+++ b/tools/perf/util/libunwind-arch/libunwind-mips.c
@@ -1,6 +1,7 @@
// SPDX-License-Identifier: GPL-2.0
#include "libunwind-arch.h"
#include "../debug.h"
+#include "../maps.h"
#include "../../../arch/mips/include/uapi/asm/perf_regs.h"
#include <linux/compiler.h>
#include <errno.h>
@@ -27,3 +28,17 @@ int __get_perf_regnum_for_unw_regnum_mips(int unw_regnum __maybe_unused)
}
#endif // HAVE_LIBUNWIND_MIPS_SUPPORT
}
+
+void __libunwind_arch__flush_access_mips(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_MIPS_SUPPORT
+ unw_flush_cache(maps__addr_space(maps), 0, 0);
+#endif
+}
+
+void __libunwind_arch__finish_access_mips(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_MIPS_SUPPORT
+ unw_destroy_addr_space(maps__addr_space(maps));
+#endif
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-ppc32.c b/tools/perf/util/libunwind-arch/libunwind-ppc32.c
index edcb0ec95dd7..fa8471c74bf3 100644
--- a/tools/perf/util/libunwind-arch/libunwind-ppc32.c
+++ b/tools/perf/util/libunwind-arch/libunwind-ppc32.c
@@ -1,6 +1,7 @@
// SPDX-License-Identifier: GPL-2.0-or-later
#include "libunwind-arch.h"
#include "../debug.h"
+#include "../maps.h"
#include "../../../arch/powerpc/include/uapi/asm/perf_regs.h"
#include <linux/compiler.h>
#include <errno.h>
@@ -29,3 +30,17 @@ int __get_perf_regnum_for_unw_regnum_ppc32(int unw_regnum __maybe_unused)
}
#endif // HAVE_LIBUNWIND_PPC32_SUPPORT
}
+
+void __libunwind_arch__flush_access_ppc32(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_PPC32_SUPPORT
+ unw_flush_cache(maps__addr_space(maps), 0, 0);
+#endif
+}
+
+void __libunwind_arch__finish_access_ppc32(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_PPC32_SUPPORT
+ unw_destroy_addr_space(maps__addr_space(maps));
+#endif
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-ppc64.c b/tools/perf/util/libunwind-arch/libunwind-ppc64.c
index 9f57a049600b..2f746e347336 100644
--- a/tools/perf/util/libunwind-arch/libunwind-ppc64.c
+++ b/tools/perf/util/libunwind-arch/libunwind-ppc64.c
@@ -1,6 +1,7 @@
// SPDX-License-Identifier: GPL-2.0-or-later
#include "libunwind-arch.h"
#include "../debug.h"
+#include "../maps.h"
#include "../../../arch/powerpc/include/uapi/asm/perf_regs.h"
#include <linux/compiler.h>
#include <errno.h>
@@ -31,3 +32,17 @@ int __get_perf_regnum_for_unw_regnum_ppc64(int unw_regnum __maybe_unused)
}
#endif // HAVE_LIBUNWIND_PPC64_SUPPORT
}
+
+void __libunwind_arch__flush_access_ppc64(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_PPC64_SUPPORT
+ unw_flush_cache(maps__addr_space(maps), 0, 0);
+#endif
+}
+
+void __libunwind_arch__finish_access_ppc64(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_PPC64_SUPPORT
+ unw_destroy_addr_space(maps__addr_space(maps));
+#endif
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-s390.c b/tools/perf/util/libunwind-arch/libunwind-s390.c
index 9fcc7885ca55..9f68d15438b2 100644
--- a/tools/perf/util/libunwind-arch/libunwind-s390.c
+++ b/tools/perf/util/libunwind-arch/libunwind-s390.c
@@ -1,6 +1,7 @@
// SPDX-License-Identifier: GPL-2.0
#include "libunwind-arch.h"
#include "../debug.h"
+#include "../maps.h"
#include "../../../arch/s390/include/uapi/asm/perf_regs.h"
#include <linux/compiler.h>
#include <errno.h>
@@ -27,3 +28,17 @@ int __get_perf_regnum_for_unw_regnum_s390(int unw_regnum __maybe_unused)
}
#endif // HAVE_LIBUNWIND_S390X_SUPPORT
}
+
+void __libunwind_arch__flush_access_s390(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_S390X_SUPPORT
+ unw_flush_cache(maps__addr_space(maps), 0, 0);
+#endif
+}
+
+void __libunwind_arch__finish_access_s390(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_S390X_SUPPORT
+ unw_destroy_addr_space(maps__addr_space(maps));
+#endif
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-x86_64.c b/tools/perf/util/libunwind-arch/libunwind-x86_64.c
index 6072e3597e61..25e326bd3e15 100644
--- a/tools/perf/util/libunwind-arch/libunwind-x86_64.c
+++ b/tools/perf/util/libunwind-arch/libunwind-x86_64.c
@@ -1,6 +1,7 @@
// SPDX-License-Identifier: GPL-2.0
#include "libunwind-arch.h"
#include "../debug.h"
+#include "../maps.h"
#include "../../../arch/x86/include/uapi/asm/perf_regs.h"
#include <linux/compiler.h>
#include <linux/kernel.h>
@@ -50,3 +51,17 @@ int __get_perf_regnum_for_unw_regnum_x86_64(int unw_regnum __maybe_unused)
return perf_x86_64_regnums[unw_regnum];
#endif // HAVE_LIBUNWIND_X86_64_SUPPORT
}
+
+void __libunwind_arch__flush_access_x86_64(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_X86_64_SUPPORT
+ unw_flush_cache(maps__addr_space(maps), 0, 0);
+#endif
+}
+
+void __libunwind_arch__finish_access_x86_64(struct maps *maps __maybe_unused)
+{
+#ifdef HAVE_LIBUNWIND_X86_64_SUPPORT
+ unw_destroy_addr_space(maps__addr_space(maps));
+#endif
+}
diff --git a/tools/perf/util/maps.c b/tools/perf/util/maps.c
index 4092211cff62..8c7b2a1e7642 100644
--- a/tools/perf/util/maps.c
+++ b/tools/perf/util/maps.c
@@ -40,6 +40,7 @@ DECLARE_RC_STRUCT(maps) {
#ifdef HAVE_LIBUNWIND_SUPPORT
void *addr_space;
const struct unwind_libunwind_ops *unwind_libunwind_ops;
+ uint16_t e_machine;
#endif
#ifdef HAVE_LIBDW_SUPPORT
void *libdw_addr_space_dwfl;
@@ -206,6 +207,16 @@ void maps__set_unwind_libunwind_ops(struct maps *maps, const struct unwind_libun
{
RC_CHK_ACCESS(maps)->unwind_libunwind_ops = ops;
}
+
+uint16_t maps__e_machine(const struct maps *maps)
+{
+ return RC_CHK_ACCESS(maps)->e_machine;
+}
+
+void maps__set_e_machine(struct maps *maps, uint16_t e_machine)
+{
+ RC_CHK_ACCESS(maps)->e_machine = e_machine;
+}
#endif
#ifdef HAVE_LIBDW_SUPPORT
void *maps__libdw_addr_space_dwfl(const struct maps *maps)
@@ -1038,6 +1049,9 @@ int maps__copy_from(struct maps *dest, struct maps *parent)
down_write(maps__lock(dest));
down_read(maps__lock(parent));
+#ifdef HAVE_LIBUNWIND_SUPPORT
+ unwind__prepare_access(dest, maps__e_machine(parent));
+#endif
parent_maps_by_address = maps__maps_by_address(parent);
n = maps__nr_maps(parent);
if (maps__nr_maps(dest) == 0) {
@@ -1067,14 +1081,11 @@ int maps__copy_from(struct maps *dest, struct maps *parent)
if (!new)
err = -ENOMEM;
else {
- err = unwind__prepare_access(dest, new, NULL);
- if (!err) {
- dest_maps_by_address[i] = new;
- map__set_kmap_maps(new, dest);
- if (dest_maps_by_name)
- dest_maps_by_name[i] = map__get(new);
- RC_CHK_ACCESS(dest)->nr_maps = i + 1;
- }
+ dest_maps_by_address[i] = new;
+ map__set_kmap_maps(new, dest);
+ if (dest_maps_by_name)
+ dest_maps_by_name[i] = map__get(new);
+ RC_CHK_ACCESS(dest)->nr_maps = i + 1;
}
if (err)
map__put(new);
@@ -1099,9 +1110,7 @@ int maps__copy_from(struct maps *dest, struct maps *parent)
if (!new)
err = -ENOMEM;
else {
- err = unwind__prepare_access(dest, new, NULL);
- if (!err)
- err = __maps__insert(dest, new);
+ err = __maps__insert(dest, new);
}
map__put(new);
}
diff --git a/tools/perf/util/maps.h b/tools/perf/util/maps.h
index 20c52084ba9e..6469f62c41a8 100644
--- a/tools/perf/util/maps.h
+++ b/tools/perf/util/maps.h
@@ -51,6 +51,8 @@ void *maps__addr_space(const struct maps *maps);
void maps__set_addr_space(struct maps *maps, void *addr_space);
const struct unwind_libunwind_ops *maps__unwind_libunwind_ops(const struct maps *maps);
void maps__set_unwind_libunwind_ops(struct maps *maps, const struct unwind_libunwind_ops *ops);
+uint16_t maps__e_machine(const struct maps *maps);
+void maps__set_e_machine(struct maps *maps, uint16_t e_machine);
#endif
#ifdef HAVE_LIBDW_SUPPORT
void *maps__libdw_addr_space_dwfl(const struct maps *maps);
diff --git a/tools/perf/util/thread.c b/tools/perf/util/thread.c
index 22be77225bb0..c5df11ad329c 100644
--- a/tools/perf/util/thread.c
+++ b/tools/perf/util/thread.c
@@ -358,41 +358,20 @@ size_t thread__fprintf(struct thread *thread, FILE *fp)
int thread__insert_map(struct thread *thread, struct map *map)
{
int ret;
+ uint16_t e_machine = thread__e_machine(thread, /*machine=*/NULL, /*e_flags=*/NULL);
- ret = unwind__prepare_access(thread__maps(thread), map, NULL);
+ ret = unwind__prepare_access(thread__maps(thread), e_machine);
if (ret)
return ret;
return maps__fixup_overlap_and_insert(thread__maps(thread), map);
}
-struct thread__prepare_access_maps_cb_args {
- int err;
- struct maps *maps;
-};
-
-static int thread__prepare_access_maps_cb(struct map *map, void *data)
-{
- bool initialized = false;
- struct thread__prepare_access_maps_cb_args *args = data;
-
- args->err = unwind__prepare_access(args->maps, map, &initialized);
-
- return (args->err || initialized) ? 1 : 0;
-}
-
static int thread__prepare_access(struct thread *thread)
{
- struct thread__prepare_access_maps_cb_args args = {
- .err = 0,
- };
-
- if (dwarf_callchain_users) {
- args.maps = thread__maps(thread);
- maps__for_each_map(thread__maps(thread), thread__prepare_access_maps_cb, &args);
- }
+ uint16_t e_machine = thread__e_machine(thread, /*machine=*/NULL, /*e_flags=*/NULL);
- return args.err;
+ return unwind__prepare_access(thread__maps(thread), e_machine);
}
static int thread__clone_maps(struct thread *thread, struct thread *parent, bool do_maps_clone)
diff --git a/tools/perf/util/unwind-libunwind-local.c b/tools/perf/util/unwind-libunwind-local.c
index 3ecdb468b859..8f0128ba05a7 100644
--- a/tools/perf/util/unwind-libunwind-local.c
+++ b/tools/perf/util/unwind-libunwind-local.c
@@ -722,16 +722,6 @@ static int _unwind__prepare_access(struct maps *maps)
return 0;
}
-static void _unwind__flush_access(struct maps *maps)
-{
- unw_flush_cache(maps__addr_space(maps), 0, 0);
-}
-
-static void _unwind__finish_access(struct maps *maps)
-{
- unw_destroy_addr_space(maps__addr_space(maps));
-}
-
static int get_entries(struct unwind_info *ui, unwind_entry_cb_t cb,
void *arg, int max_stack)
{
@@ -821,8 +811,6 @@ static int _unwind__get_entries(unwind_entry_cb_t cb, void *arg,
static struct unwind_libunwind_ops
_unwind_libunwind_ops = {
.prepare_access = _unwind__prepare_access,
- .flush_access = _unwind__flush_access,
- .finish_access = _unwind__finish_access,
.get_entries = _unwind__get_entries,
};
diff --git a/tools/perf/util/unwind-libunwind.c b/tools/perf/util/unwind-libunwind.c
index a0016b897dae..eaee7b78d87c 100644
--- a/tools/perf/util/unwind-libunwind.c
+++ b/tools/perf/util/unwind-libunwind.c
@@ -7,76 +7,55 @@
#include "debug.h"
#include "env.h"
#include "callchain.h"
+#include "libunwind-arch/libunwind-arch.h"
+#include <dwarf-regs.h>
+#include <elf.h>
struct unwind_libunwind_ops __weak *local_unwind_libunwind_ops;
struct unwind_libunwind_ops __weak *x86_32_unwind_libunwind_ops;
struct unwind_libunwind_ops __weak *arm64_unwind_libunwind_ops;
-int unwind__prepare_access(struct maps *maps, struct map *map, bool *initialized)
+int unwind__prepare_access(struct maps *maps, uint16_t e_machine)
{
- const char *arch;
- enum dso_type dso_type;
struct unwind_libunwind_ops *ops = local_unwind_libunwind_ops;
- struct dso *dso = map__dso(map);
- struct machine *machine;
- int err;
if (!dwarf_callchain_users)
return 0;
if (maps__addr_space(maps)) {
- pr_debug("unwind: thread map already set, dso=%s\n", dso__name(dso));
- if (initialized)
- *initialized = true;
+ pr_debug3("unwind: thread map already set\n");
return 0;
}
- machine = maps__machine(maps);
- /* env->arch is NULL for live-mode (i.e. perf top) */
- if (!machine->env || !machine->env->arch)
- goto out_register;
-
- dso_type = dso__type(dso, machine);
- if (dso_type == DSO__TYPE_UNKNOWN)
- return 0;
-
- arch = perf_env__arch(machine->env);
-
- if (!strcmp(arch, "x86")) {
- if (dso_type != DSO__TYPE_64BIT)
+ if (e_machine != EM_HOST) {
+ /* If not live/local mode. */
+ switch (e_machine) {
+ case EM_386:
ops = x86_32_unwind_libunwind_ops;
- } else if (!strcmp(arch, "arm64") || !strcmp(arch, "arm")) {
- if (dso_type == DSO__TYPE_64BIT)
+ break;
+ case EM_AARCH64:
ops = arm64_unwind_libunwind_ops;
- }
-
- if (!ops) {
- pr_warning_once("unwind: target platform=%s is not supported\n", arch);
+ break;
+ default:
+ pr_warning_once("unwind: ELF machine type %d is not supported\n",
+ e_machine);
return 0;
+ }
}
-out_register:
maps__set_unwind_libunwind_ops(maps, ops);
+ maps__set_e_machine(maps, e_machine);
- err = maps__unwind_libunwind_ops(maps)->prepare_access(maps);
- if (initialized)
- *initialized = err ? false : true;
- return err;
+ return maps__unwind_libunwind_ops(maps)->prepare_access(maps);
}
void unwind__flush_access(struct maps *maps)
{
- const struct unwind_libunwind_ops *ops = maps__unwind_libunwind_ops(maps);
-
- if (ops)
- ops->flush_access(maps);
+ libunwind_arch__flush_access(maps);
}
void unwind__finish_access(struct maps *maps)
{
- const struct unwind_libunwind_ops *ops = maps__unwind_libunwind_ops(maps);
-
- if (ops)
- ops->finish_access(maps);
+ libunwind_arch__finish_access(maps);
}
int libunwind__get_entries(unwind_entry_cb_t cb, void *arg,
diff --git a/tools/perf/util/unwind.h b/tools/perf/util/unwind.h
index f8a8788ac986..ad610c5a241c 100644
--- a/tools/perf/util/unwind.h
+++ b/tools/perf/util/unwind.h
@@ -2,6 +2,7 @@
#ifndef __UNWIND_H
#define __UNWIND_H
+#include <stdint.h>
#include <linux/compiler.h>
#include <linux/types.h>
#include "map_symbol.h"
@@ -20,8 +21,6 @@ typedef int (*unwind_entry_cb_t)(struct unwind_entry *entry, void *arg);
struct unwind_libunwind_ops {
int (*prepare_access)(struct maps *maps);
- void (*flush_access)(struct maps *maps);
- void (*finish_access)(struct maps *maps);
int (*get_entries)(unwind_entry_cb_t cb, void *arg,
struct thread *thread,
struct perf_sample *data, int max_stack, bool best_effort);
@@ -64,7 +63,7 @@ int libunwind__get_entries(unwind_entry_cb_t cb, void *arg,
struct thread *thread,
struct perf_sample *data, int max_stack,
bool best_effort);
-int unwind__prepare_access(struct maps *maps, struct map *map, bool *initialized);
+int unwind__prepare_access(struct maps *maps, uint16_t e_machine);
void unwind__flush_access(struct maps *maps);
void unwind__finish_access(struct maps *maps);
#else
@@ -81,8 +80,7 @@ static inline int libunwind__get_entries(unwind_entry_cb_t cb __maybe_unused,
}
static inline int unwind__prepare_access(struct maps *maps __maybe_unused,
- struct map *map __maybe_unused,
- bool *initialized __maybe_unused)
+ uint16_t e_machine __maybe_unused)
{
return 0;
}
--
2.53.0.1213.gd9a14994de-goog
^ permalink raw reply related
* [PATCH v3 5/8] perf unwind-libunwind: Make libunwind register reading cross platform
From: Ian Rogers @ 2026-04-04 5:40 UTC (permalink / raw)
To: acme
Cc: irogers, 9erthalion6, adrian.hunter, alex, alexander.shishkin,
andrew.jones, aou, atrajeev, blakejones, ctshao, dapeng1.mi,
howardchu95, james.clark, john.g.garry, jolsa, leo.yan,
libunwind-devel, linux-arm-kernel, linux-kernel, linux-perf-users,
linux-riscv, mingo, namhyung, palmer, peterz, pjw, shimin.guo,
tglozar, tmricht, will, yuzhuo
In-Reply-To: <20260404054032.1538095-1-irogers@google.com>
Move the libunwind register to perf register mapping functions in
arch/../util/unwind-libunwind.c into a new libunwind-arch
directory. Rename the functions to
__get_perf_regnum_for_unw_regnum_<arch>. Add untested ppc32 and s390
functions. Add a get_perf_regnum_for_unw_regnum function that takes an
ELF machine as well as a register number and chooses the appropriate
architecture implementation.
Split the x86 and powerpc 32 and 64-bit implementations apart so that
a single libunwind-<arch>.h header is included.
Move the e_machine into the unwind_info struct to make it easier to
pass.
Signed-off-by: Ian Rogers <irogers@google.com>
---
tools/perf/arch/arm/util/Build | 2 -
tools/perf/arch/arm/util/unwind-libunwind.c | 50 --------
tools/perf/arch/arm64/util/Build | 1 -
tools/perf/arch/arm64/util/unwind-libunwind.c | 17 ---
tools/perf/arch/loongarch/util/Build | 2 -
.../arch/loongarch/util/unwind-libunwind.c | 82 -------------
tools/perf/arch/mips/Build | 1 -
tools/perf/arch/mips/util/Build | 1 -
tools/perf/arch/mips/util/unwind-libunwind.c | 22 ----
tools/perf/arch/powerpc/util/Build | 1 -
.../perf/arch/powerpc/util/unwind-libunwind.c | 92 --------------
tools/perf/arch/x86/util/Build | 3 -
tools/perf/arch/x86/util/unwind-libunwind.c | 115 ------------------
tools/perf/util/Build | 1 +
tools/perf/util/libunwind-arch/Build | 10 ++
.../perf/util/libunwind-arch/libunwind-arch.c | 32 +++++
.../perf/util/libunwind-arch/libunwind-arch.h | 16 +++
.../perf/util/libunwind-arch/libunwind-arm.c | 15 +++
.../util/libunwind-arch/libunwind-arm64.c | 14 +++
.../perf/util/libunwind-arch/libunwind-i386.c | 43 +++++++
.../util/libunwind-arch/libunwind-loongarch.c | 27 ++++
.../perf/util/libunwind-arch/libunwind-mips.c | 29 +++++
.../util/libunwind-arch/libunwind-ppc32.c | 31 +++++
.../util/libunwind-arch/libunwind-ppc64.c | 33 +++++
.../perf/util/libunwind-arch/libunwind-s390.c | 29 +++++
.../util/libunwind-arch/libunwind-x86_64.c | 52 ++++++++
tools/perf/util/libunwind/arm64.c | 5 -
tools/perf/util/libunwind/x86_32.c | 12 --
tools/perf/util/unwind-libunwind-local.c | 12 +-
tools/perf/util/unwind.h | 5 -
30 files changed, 338 insertions(+), 417 deletions(-)
delete mode 100644 tools/perf/arch/arm/util/unwind-libunwind.c
delete mode 100644 tools/perf/arch/arm64/util/unwind-libunwind.c
delete mode 100644 tools/perf/arch/loongarch/util/unwind-libunwind.c
delete mode 100644 tools/perf/arch/mips/Build
delete mode 100644 tools/perf/arch/mips/util/Build
delete mode 100644 tools/perf/arch/mips/util/unwind-libunwind.c
delete mode 100644 tools/perf/arch/powerpc/util/unwind-libunwind.c
delete mode 100644 tools/perf/arch/x86/util/unwind-libunwind.c
create mode 100644 tools/perf/util/libunwind-arch/Build
create mode 100644 tools/perf/util/libunwind-arch/libunwind-arch.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-arch.h
create mode 100644 tools/perf/util/libunwind-arch/libunwind-arm.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-arm64.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-i386.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-loongarch.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-mips.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-ppc32.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-ppc64.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-s390.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-x86_64.c
diff --git a/tools/perf/arch/arm/util/Build b/tools/perf/arch/arm/util/Build
index b94bf3c5279a..768ae5d16553 100644
--- a/tools/perf/arch/arm/util/Build
+++ b/tools/perf/arch/arm/util/Build
@@ -1,3 +1 @@
-perf-util-$(CONFIG_LOCAL_LIBUNWIND) += unwind-libunwind.o
-
perf-util-y += pmu.o auxtrace.o cs-etm.o
diff --git a/tools/perf/arch/arm/util/unwind-libunwind.c b/tools/perf/arch/arm/util/unwind-libunwind.c
deleted file mode 100644
index 438906bf0014..000000000000
--- a/tools/perf/arch/arm/util/unwind-libunwind.c
+++ /dev/null
@@ -1,50 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-
-#include <errno.h>
-#include <libunwind.h>
-#include "perf_regs.h"
-#include "../../../util/unwind.h"
-#include "../../../util/debug.h"
-
-int libunwind__arch_reg_id(int regnum)
-{
- switch (regnum) {
- case UNW_ARM_R0:
- return PERF_REG_ARM_R0;
- case UNW_ARM_R1:
- return PERF_REG_ARM_R1;
- case UNW_ARM_R2:
- return PERF_REG_ARM_R2;
- case UNW_ARM_R3:
- return PERF_REG_ARM_R3;
- case UNW_ARM_R4:
- return PERF_REG_ARM_R4;
- case UNW_ARM_R5:
- return PERF_REG_ARM_R5;
- case UNW_ARM_R6:
- return PERF_REG_ARM_R6;
- case UNW_ARM_R7:
- return PERF_REG_ARM_R7;
- case UNW_ARM_R8:
- return PERF_REG_ARM_R8;
- case UNW_ARM_R9:
- return PERF_REG_ARM_R9;
- case UNW_ARM_R10:
- return PERF_REG_ARM_R10;
- case UNW_ARM_R11:
- return PERF_REG_ARM_FP;
- case UNW_ARM_R12:
- return PERF_REG_ARM_IP;
- case UNW_ARM_R13:
- return PERF_REG_ARM_SP;
- case UNW_ARM_R14:
- return PERF_REG_ARM_LR;
- case UNW_ARM_R15:
- return PERF_REG_ARM_PC;
- default:
- pr_err("unwind: invalid reg id %d\n", regnum);
- return -EINVAL;
- }
-
- return -EINVAL;
-}
diff --git a/tools/perf/arch/arm64/util/Build b/tools/perf/arch/arm64/util/Build
index 4e06a08d281a..4b70c4788c80 100644
--- a/tools/perf/arch/arm64/util/Build
+++ b/tools/perf/arch/arm64/util/Build
@@ -1,4 +1,3 @@
-perf-util-$(CONFIG_LOCAL_LIBUNWIND) += unwind-libunwind.o
perf-util-y += ../../arm/util/auxtrace.o
perf-util-y += ../../arm/util/cs-etm.o
perf-util-y += ../../arm/util/pmu.o
diff --git a/tools/perf/arch/arm64/util/unwind-libunwind.c b/tools/perf/arch/arm64/util/unwind-libunwind.c
deleted file mode 100644
index 871af5992298..000000000000
--- a/tools/perf/arch/arm64/util/unwind-libunwind.c
+++ /dev/null
@@ -1,17 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-#include <errno.h>
-
-#ifndef REMOTE_UNWIND_LIBUNWIND
-#include <libunwind.h>
-#include "perf_regs.h"
-#include "../../../util/unwind.h"
-#endif
-#include "../../../util/debug.h"
-
-int LIBUNWIND__ARCH_REG_ID(int regnum)
-{
- if (regnum < 0 || regnum >= PERF_REG_ARM64_EXTENDED_MAX)
- return -EINVAL;
-
- return regnum;
-}
diff --git a/tools/perf/arch/loongarch/util/Build b/tools/perf/arch/loongarch/util/Build
index 8d91e78d31c9..2328fb9a30a3 100644
--- a/tools/perf/arch/loongarch/util/Build
+++ b/tools/perf/arch/loongarch/util/Build
@@ -1,3 +1 @@
perf-util-y += header.o
-
-perf-util-$(CONFIG_LOCAL_LIBUNWIND) += unwind-libunwind.o
diff --git a/tools/perf/arch/loongarch/util/unwind-libunwind.c b/tools/perf/arch/loongarch/util/unwind-libunwind.c
deleted file mode 100644
index f693167b86ef..000000000000
--- a/tools/perf/arch/loongarch/util/unwind-libunwind.c
+++ /dev/null
@@ -1,82 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-
-#include <errno.h>
-#include <libunwind.h>
-#include "perf_regs.h"
-#include "../../util/unwind.h"
-#include "util/debug.h"
-
-int libunwind__arch_reg_id(int regnum)
-{
- switch (regnum) {
- case UNW_LOONGARCH64_R1:
- return PERF_REG_LOONGARCH_R1;
- case UNW_LOONGARCH64_R2:
- return PERF_REG_LOONGARCH_R2;
- case UNW_LOONGARCH64_R3:
- return PERF_REG_LOONGARCH_R3;
- case UNW_LOONGARCH64_R4:
- return PERF_REG_LOONGARCH_R4;
- case UNW_LOONGARCH64_R5:
- return PERF_REG_LOONGARCH_R5;
- case UNW_LOONGARCH64_R6:
- return PERF_REG_LOONGARCH_R6;
- case UNW_LOONGARCH64_R7:
- return PERF_REG_LOONGARCH_R7;
- case UNW_LOONGARCH64_R8:
- return PERF_REG_LOONGARCH_R8;
- case UNW_LOONGARCH64_R9:
- return PERF_REG_LOONGARCH_R9;
- case UNW_LOONGARCH64_R10:
- return PERF_REG_LOONGARCH_R10;
- case UNW_LOONGARCH64_R11:
- return PERF_REG_LOONGARCH_R11;
- case UNW_LOONGARCH64_R12:
- return PERF_REG_LOONGARCH_R12;
- case UNW_LOONGARCH64_R13:
- return PERF_REG_LOONGARCH_R13;
- case UNW_LOONGARCH64_R14:
- return PERF_REG_LOONGARCH_R14;
- case UNW_LOONGARCH64_R15:
- return PERF_REG_LOONGARCH_R15;
- case UNW_LOONGARCH64_R16:
- return PERF_REG_LOONGARCH_R16;
- case UNW_LOONGARCH64_R17:
- return PERF_REG_LOONGARCH_R17;
- case UNW_LOONGARCH64_R18:
- return PERF_REG_LOONGARCH_R18;
- case UNW_LOONGARCH64_R19:
- return PERF_REG_LOONGARCH_R19;
- case UNW_LOONGARCH64_R20:
- return PERF_REG_LOONGARCH_R20;
- case UNW_LOONGARCH64_R21:
- return PERF_REG_LOONGARCH_R21;
- case UNW_LOONGARCH64_R22:
- return PERF_REG_LOONGARCH_R22;
- case UNW_LOONGARCH64_R23:
- return PERF_REG_LOONGARCH_R23;
- case UNW_LOONGARCH64_R24:
- return PERF_REG_LOONGARCH_R24;
- case UNW_LOONGARCH64_R25:
- return PERF_REG_LOONGARCH_R25;
- case UNW_LOONGARCH64_R26:
- return PERF_REG_LOONGARCH_R26;
- case UNW_LOONGARCH64_R27:
- return PERF_REG_LOONGARCH_R27;
- case UNW_LOONGARCH64_R28:
- return PERF_REG_LOONGARCH_R28;
- case UNW_LOONGARCH64_R29:
- return PERF_REG_LOONGARCH_R29;
- case UNW_LOONGARCH64_R30:
- return PERF_REG_LOONGARCH_R30;
- case UNW_LOONGARCH64_R31:
- return PERF_REG_LOONGARCH_R31;
- case UNW_LOONGARCH64_PC:
- return PERF_REG_LOONGARCH_PC;
- default:
- pr_err("unwind: invalid reg id %d\n", regnum);
- return -EINVAL;
- }
-
- return -EINVAL;
-}
diff --git a/tools/perf/arch/mips/Build b/tools/perf/arch/mips/Build
deleted file mode 100644
index e63eabc2c8f4..000000000000
--- a/tools/perf/arch/mips/Build
+++ /dev/null
@@ -1 +0,0 @@
-perf-util-y += util/
diff --git a/tools/perf/arch/mips/util/Build b/tools/perf/arch/mips/util/Build
deleted file mode 100644
index 818b808a8247..000000000000
--- a/tools/perf/arch/mips/util/Build
+++ /dev/null
@@ -1 +0,0 @@
-perf-util-$(CONFIG_LOCAL_LIBUNWIND) += unwind-libunwind.o
diff --git a/tools/perf/arch/mips/util/unwind-libunwind.c b/tools/perf/arch/mips/util/unwind-libunwind.c
deleted file mode 100644
index 0d8c99c29da6..000000000000
--- a/tools/perf/arch/mips/util/unwind-libunwind.c
+++ /dev/null
@@ -1,22 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-
-#include <errno.h>
-#include <libunwind.h>
-#include "perf_regs.h"
-#include "../../util/unwind.h"
-#include "util/debug.h"
-
-int libunwind__arch_reg_id(int regnum)
-{
- switch (regnum) {
- case UNW_MIPS_R1 ... UNW_MIPS_R25:
- return regnum - UNW_MIPS_R1 + PERF_REG_MIPS_R1;
- case UNW_MIPS_R28 ... UNW_MIPS_R31:
- return regnum - UNW_MIPS_R28 + PERF_REG_MIPS_R28;
- case UNW_MIPS_PC:
- return PERF_REG_MIPS_PC;
- default:
- pr_err("unwind: invalid reg id %d\n", regnum);
- return -EINVAL;
- }
-}
diff --git a/tools/perf/arch/powerpc/util/Build b/tools/perf/arch/powerpc/util/Build
index d66574cbb9a9..ae928050e07a 100644
--- a/tools/perf/arch/powerpc/util/Build
+++ b/tools/perf/arch/powerpc/util/Build
@@ -6,5 +6,4 @@ perf-util-y += evsel.o
perf-util-$(CONFIG_LIBDW) += skip-callchain-idx.o
-perf-util-$(CONFIG_LIBUNWIND) += unwind-libunwind.o
perf-util-y += auxtrace.o
diff --git a/tools/perf/arch/powerpc/util/unwind-libunwind.c b/tools/perf/arch/powerpc/util/unwind-libunwind.c
deleted file mode 100644
index 90a6beda20de..000000000000
--- a/tools/perf/arch/powerpc/util/unwind-libunwind.c
+++ /dev/null
@@ -1,92 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0-or-later
-/*
- * Copyright 2016 Chandan Kumar, IBM Corporation.
- */
-
-#include <errno.h>
-#include <libunwind.h>
-#include <asm/perf_regs.h>
-#include "../../util/unwind.h"
-#include "../../util/debug.h"
-
-int libunwind__arch_reg_id(int regnum)
-{
- switch (regnum) {
- case UNW_PPC64_R0:
- return PERF_REG_POWERPC_R0;
- case UNW_PPC64_R1:
- return PERF_REG_POWERPC_R1;
- case UNW_PPC64_R2:
- return PERF_REG_POWERPC_R2;
- case UNW_PPC64_R3:
- return PERF_REG_POWERPC_R3;
- case UNW_PPC64_R4:
- return PERF_REG_POWERPC_R4;
- case UNW_PPC64_R5:
- return PERF_REG_POWERPC_R5;
- case UNW_PPC64_R6:
- return PERF_REG_POWERPC_R6;
- case UNW_PPC64_R7:
- return PERF_REG_POWERPC_R7;
- case UNW_PPC64_R8:
- return PERF_REG_POWERPC_R8;
- case UNW_PPC64_R9:
- return PERF_REG_POWERPC_R9;
- case UNW_PPC64_R10:
- return PERF_REG_POWERPC_R10;
- case UNW_PPC64_R11:
- return PERF_REG_POWERPC_R11;
- case UNW_PPC64_R12:
- return PERF_REG_POWERPC_R12;
- case UNW_PPC64_R13:
- return PERF_REG_POWERPC_R13;
- case UNW_PPC64_R14:
- return PERF_REG_POWERPC_R14;
- case UNW_PPC64_R15:
- return PERF_REG_POWERPC_R15;
- case UNW_PPC64_R16:
- return PERF_REG_POWERPC_R16;
- case UNW_PPC64_R17:
- return PERF_REG_POWERPC_R17;
- case UNW_PPC64_R18:
- return PERF_REG_POWERPC_R18;
- case UNW_PPC64_R19:
- return PERF_REG_POWERPC_R19;
- case UNW_PPC64_R20:
- return PERF_REG_POWERPC_R20;
- case UNW_PPC64_R21:
- return PERF_REG_POWERPC_R21;
- case UNW_PPC64_R22:
- return PERF_REG_POWERPC_R22;
- case UNW_PPC64_R23:
- return PERF_REG_POWERPC_R23;
- case UNW_PPC64_R24:
- return PERF_REG_POWERPC_R24;
- case UNW_PPC64_R25:
- return PERF_REG_POWERPC_R25;
- case UNW_PPC64_R26:
- return PERF_REG_POWERPC_R26;
- case UNW_PPC64_R27:
- return PERF_REG_POWERPC_R27;
- case UNW_PPC64_R28:
- return PERF_REG_POWERPC_R28;
- case UNW_PPC64_R29:
- return PERF_REG_POWERPC_R29;
- case UNW_PPC64_R30:
- return PERF_REG_POWERPC_R30;
- case UNW_PPC64_R31:
- return PERF_REG_POWERPC_R31;
- case UNW_PPC64_LR:
- return PERF_REG_POWERPC_LINK;
- case UNW_PPC64_CTR:
- return PERF_REG_POWERPC_CTR;
- case UNW_PPC64_XER:
- return PERF_REG_POWERPC_XER;
- case UNW_PPC64_NIP:
- return PERF_REG_POWERPC_NIP;
- default:
- pr_err("unwind: invalid reg id %d\n", regnum);
- return -EINVAL;
- }
- return -EINVAL;
-}
diff --git a/tools/perf/arch/x86/util/Build b/tools/perf/arch/x86/util/Build
index b94c91984c66..7f89fffe4615 100644
--- a/tools/perf/arch/x86/util/Build
+++ b/tools/perf/arch/x86/util/Build
@@ -8,9 +8,6 @@ perf-util-y += evlist.o
perf-util-y += mem-events.o
perf-util-y += evsel.o
perf-util-y += iostat.o
-
-perf-util-$(CONFIG_LOCAL_LIBUNWIND) += unwind-libunwind.o
-
perf-util-y += auxtrace.o
perf-util-y += intel-pt.o
perf-util-y += intel-bts.o
diff --git a/tools/perf/arch/x86/util/unwind-libunwind.c b/tools/perf/arch/x86/util/unwind-libunwind.c
deleted file mode 100644
index 47357973b55b..000000000000
--- a/tools/perf/arch/x86/util/unwind-libunwind.c
+++ /dev/null
@@ -1,115 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-
-#include <errno.h>
-#include "../../util/debug.h"
-#ifndef REMOTE_UNWIND_LIBUNWIND
-#include <libunwind.h>
-#include "perf_regs.h"
-#include "../../util/unwind.h"
-#endif
-
-#ifdef HAVE_ARCH_X86_64_SUPPORT
-int LIBUNWIND__ARCH_REG_ID(int regnum)
-{
- int id;
-
- switch (regnum) {
- case UNW_X86_64_RAX:
- id = PERF_REG_X86_AX;
- break;
- case UNW_X86_64_RDX:
- id = PERF_REG_X86_DX;
- break;
- case UNW_X86_64_RCX:
- id = PERF_REG_X86_CX;
- break;
- case UNW_X86_64_RBX:
- id = PERF_REG_X86_BX;
- break;
- case UNW_X86_64_RSI:
- id = PERF_REG_X86_SI;
- break;
- case UNW_X86_64_RDI:
- id = PERF_REG_X86_DI;
- break;
- case UNW_X86_64_RBP:
- id = PERF_REG_X86_BP;
- break;
- case UNW_X86_64_RSP:
- id = PERF_REG_X86_SP;
- break;
- case UNW_X86_64_R8:
- id = PERF_REG_X86_R8;
- break;
- case UNW_X86_64_R9:
- id = PERF_REG_X86_R9;
- break;
- case UNW_X86_64_R10:
- id = PERF_REG_X86_R10;
- break;
- case UNW_X86_64_R11:
- id = PERF_REG_X86_R11;
- break;
- case UNW_X86_64_R12:
- id = PERF_REG_X86_R12;
- break;
- case UNW_X86_64_R13:
- id = PERF_REG_X86_R13;
- break;
- case UNW_X86_64_R14:
- id = PERF_REG_X86_R14;
- break;
- case UNW_X86_64_R15:
- id = PERF_REG_X86_R15;
- break;
- case UNW_X86_64_RIP:
- id = PERF_REG_X86_IP;
- break;
- default:
- pr_err("unwind: invalid reg id %d\n", regnum);
- return -EINVAL;
- }
-
- return id;
-}
-#else
-int LIBUNWIND__ARCH_REG_ID(int regnum)
-{
- int id;
-
- switch (regnum) {
- case UNW_X86_EAX:
- id = PERF_REG_X86_AX;
- break;
- case UNW_X86_EDX:
- id = PERF_REG_X86_DX;
- break;
- case UNW_X86_ECX:
- id = PERF_REG_X86_CX;
- break;
- case UNW_X86_EBX:
- id = PERF_REG_X86_BX;
- break;
- case UNW_X86_ESI:
- id = PERF_REG_X86_SI;
- break;
- case UNW_X86_EDI:
- id = PERF_REG_X86_DI;
- break;
- case UNW_X86_EBP:
- id = PERF_REG_X86_BP;
- break;
- case UNW_X86_ESP:
- id = PERF_REG_X86_SP;
- break;
- case UNW_X86_EIP:
- id = PERF_REG_X86_IP;
- break;
- default:
- pr_err("unwind: invalid reg id %d\n", regnum);
- return -EINVAL;
- }
-
- return id;
-}
-#endif /* HAVE_ARCH_X86_64_SUPPORT */
diff --git a/tools/perf/util/Build b/tools/perf/util/Build
index 01edfccebb88..bf4204135ccb 100644
--- a/tools/perf/util/Build
+++ b/tools/perf/util/Build
@@ -228,6 +228,7 @@ perf-util-$(CONFIG_LIBDW) += unwind-libdw.o
perf-util-$(CONFIG_LOCAL_LIBUNWIND) += unwind-libunwind-local.o
perf-util-$(CONFIG_LIBUNWIND) += unwind-libunwind.o
+perf-util-$(CONFIG_LIBUNWIND) += libunwind-arch/
perf-util-$(CONFIG_LIBUNWIND_X86) += libunwind/x86_32.o
perf-util-$(CONFIG_LIBUNWIND_AARCH64) += libunwind/arm64.o
diff --git a/tools/perf/util/libunwind-arch/Build b/tools/perf/util/libunwind-arch/Build
new file mode 100644
index 000000000000..87fd657a3248
--- /dev/null
+++ b/tools/perf/util/libunwind-arch/Build
@@ -0,0 +1,10 @@
+perf-util-$(CONFIG_LIBUNWIND) += libunwind-arch.o
+perf-util-$(CONFIG_LIBUNWIND) += libunwind-arm64.o
+perf-util-$(CONFIG_LIBUNWIND) += libunwind-arm.o
+perf-util-$(CONFIG_LIBUNWIND) += libunwind-loongarch.o
+perf-util-$(CONFIG_LIBUNWIND) += libunwind-mips.o
+perf-util-$(CONFIG_LIBUNWIND) += libunwind-ppc32.o
+perf-util-$(CONFIG_LIBUNWIND) += libunwind-ppc64.o
+perf-util-$(CONFIG_LIBUNWIND) += libunwind-s390.o
+perf-util-$(CONFIG_LIBUNWIND) += libunwind-i386.o
+perf-util-$(CONFIG_LIBUNWIND) += libunwind-x86_64.o
diff --git a/tools/perf/util/libunwind-arch/libunwind-arch.c b/tools/perf/util/libunwind-arch/libunwind-arch.c
new file mode 100644
index 000000000000..5439bf90d161
--- /dev/null
+++ b/tools/perf/util/libunwind-arch/libunwind-arch.c
@@ -0,0 +1,32 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "libunwind-arch.h"
+#include "../debug.h"
+#include <elf.h>
+#include <errno.h>
+
+int get_perf_regnum_for_unw_regnum(unsigned int e_machine, int unw_regnum)
+{
+ switch (e_machine) {
+ case EM_ARM:
+ return __get_perf_regnum_for_unw_regnum_arm(unw_regnum);
+ case EM_AARCH64:
+ return __get_perf_regnum_for_unw_regnum_arm64(unw_regnum);
+ case EM_LOONGARCH:
+ return __get_perf_regnum_for_unw_regnum_loongarch(unw_regnum);
+ case EM_MIPS:
+ return __get_perf_regnum_for_unw_regnum_mips(unw_regnum);
+ case EM_PPC:
+ return __get_perf_regnum_for_unw_regnum_ppc32(unw_regnum);
+ case EM_PPC64:
+ return __get_perf_regnum_for_unw_regnum_ppc64(unw_regnum);
+ case EM_S390:
+ return __get_perf_regnum_for_unw_regnum_s390(unw_regnum);
+ case EM_386:
+ return __get_perf_regnum_for_unw_regnum_i386(unw_regnum);
+ case EM_X86_64:
+ return __get_perf_regnum_for_unw_regnum_x86_64(unw_regnum);
+ default:
+ pr_err("ELF MACHINE %x is not supported.\n", e_machine);
+ return -EINVAL;
+ }
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-arch.h b/tools/perf/util/libunwind-arch/libunwind-arch.h
new file mode 100644
index 000000000000..e1009c6cb965
--- /dev/null
+++ b/tools/perf/util/libunwind-arch/libunwind-arch.h
@@ -0,0 +1,16 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LIBUNWIND_ARCH_H
+#define __LIBUNWIND_ARCH_H
+
+int __get_perf_regnum_for_unw_regnum_arm(int unw_regnum);
+int __get_perf_regnum_for_unw_regnum_arm64(int unw_regnum);
+int __get_perf_regnum_for_unw_regnum_loongarch(int unw_regnum);
+int __get_perf_regnum_for_unw_regnum_mips(int unw_regnum);
+int __get_perf_regnum_for_unw_regnum_ppc32(int unw_regnum);
+int __get_perf_regnum_for_unw_regnum_ppc64(int unw_regnum);
+int __get_perf_regnum_for_unw_regnum_s390(int unw_regnum);
+int __get_perf_regnum_for_unw_regnum_i386(int unw_regnum);
+int __get_perf_regnum_for_unw_regnum_x86_64(int unw_regnum);
+int get_perf_regnum_for_unw_regnum(unsigned int e_machine, int unw_regnum);
+
+#endif /* __LIBUNWIND_ARCH_H */
diff --git a/tools/perf/util/libunwind-arch/libunwind-arm.c b/tools/perf/util/libunwind-arch/libunwind-arm.c
new file mode 100644
index 000000000000..6740ee55b043
--- /dev/null
+++ b/tools/perf/util/libunwind-arch/libunwind-arm.c
@@ -0,0 +1,15 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "libunwind-arch.h"
+#include "../debug.h"
+#include "../../../arch/arm/include/uapi/asm/perf_regs.h"
+#include <linux/compiler.h>
+#include <errno.h>
+
+int __get_perf_regnum_for_unw_regnum_arm(int unw_regnum)
+{
+ if (unw_regnum < 0 || unw_regnum >= PERF_REG_ARM_MAX) {
+ pr_err("unwind: invalid reg id %d\n", unw_regnum);
+ return -EINVAL;
+ }
+ return unw_regnum;
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-arm64.c b/tools/perf/util/libunwind-arch/libunwind-arm64.c
new file mode 100644
index 000000000000..53b1877dfa04
--- /dev/null
+++ b/tools/perf/util/libunwind-arch/libunwind-arm64.c
@@ -0,0 +1,14 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "libunwind-arch.h"
+#include "../debug.h"
+#include "../../../arch/arm64/include/uapi/asm/perf_regs.h"
+#include <errno.h>
+
+int __get_perf_regnum_for_unw_regnum_arm64(int unw_regnum)
+{
+ if (unw_regnum < 0 || unw_regnum >= PERF_REG_ARM64_EXTENDED_MAX) {
+ pr_err("unwind: invalid reg id %d\n", unw_regnum);
+ return -EINVAL;
+ }
+ return unw_regnum;
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-i386.c b/tools/perf/util/libunwind-arch/libunwind-i386.c
new file mode 100644
index 000000000000..9eaf4ebff0c2
--- /dev/null
+++ b/tools/perf/util/libunwind-arch/libunwind-i386.c
@@ -0,0 +1,43 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "libunwind-arch.h"
+#include "../debug.h"
+#include "../../../arch/x86/include/uapi/asm/perf_regs.h"
+#include <linux/compiler.h>
+#include <linux/kernel.h>
+#include <errno.h>
+
+#ifdef HAVE_LIBUNWIND_X86_SUPPORT
+#include <libunwind-x86.h>
+#endif
+
+int __get_perf_regnum_for_unw_regnum_i386(int unw_regnum __maybe_unused)
+{
+#ifndef HAVE_LIBUNWIND_X86_SUPPORT
+ return -EINVAL;
+#else
+ static const int perf_i386_regnums[] = {
+#define REGNUM(reg) [UNW_X86_E ## reg] = PERF_REG_X86_ ## reg
+ REGNUM(AX),
+ REGNUM(DX),
+ REGNUM(CX),
+ REGNUM(BX),
+ REGNUM(SI),
+ REGNUM(DI),
+ REGNUM(BP),
+ REGNUM(SP),
+ REGNUM(IP),
+#undef REGNUM
+ };
+
+ if (unw_regnum == UNW_X86_EAX)
+ return PERF_REG_X86_AX;
+
+ if (unw_regnum < 0 || unw_regnum > (int)ARRAY_SIZE(perf_i386_regnums) ||
+ perf_i386_regnums[unw_regnum] == 0) {
+ pr_err("unwind: invalid reg id %d\n", unw_regnum);
+ return -EINVAL;
+ }
+
+ return perf_i386_regnums[unw_regnum];
+#endif // HAVE_LIBUNWIND_X86_SUPPORT
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-loongarch.c b/tools/perf/util/libunwind-arch/libunwind-loongarch.c
new file mode 100644
index 000000000000..7009410989bc
--- /dev/null
+++ b/tools/perf/util/libunwind-arch/libunwind-loongarch.c
@@ -0,0 +1,27 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "libunwind-arch.h"
+#include "../debug.h"
+#include "../../../arch/loongarch/include/uapi/asm/perf_regs.h"
+#include <linux/compiler.h>
+#include <errno.h>
+
+#ifdef HAVE_LIBUNWIND_LOONGARCH64_SUPPORT
+#include <libunwind-loongarch64.h>
+#endif
+
+int __get_perf_regnum_for_unw_regnum_loongarch(int unw_regnum __maybe_unused)
+{
+#ifndef HAVE_LIBUNWIND_LOONGARCH64_SUPPORT
+ return -EINVAL;
+#else
+ switch (unw_regnum) {
+ case UNW_LOONGARCH64_R1 ... UNW_LOONGARCH64_31:
+ return unw_regnum - UNW_LOONGARCH64_R1 + PERF_REG_LOONGARCH_R1;
+ case UNW_LOONGARCH64_PC:
+ return PERF_REG_LOONGARCH_PC;
+ default:
+ pr_err("unwind: invalid reg id %d\n", unw_regnum);
+ return -EINVAL;
+ }
+#endif // HAVE_LIBUNWIND_LOONGARCH64_SUPPORT
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-mips.c b/tools/perf/util/libunwind-arch/libunwind-mips.c
new file mode 100644
index 000000000000..01a506c8079c
--- /dev/null
+++ b/tools/perf/util/libunwind-arch/libunwind-mips.c
@@ -0,0 +1,29 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "libunwind-arch.h"
+#include "../debug.h"
+#include "../../../arch/mips/include/uapi/asm/perf_regs.h"
+#include <linux/compiler.h>
+#include <errno.h>
+
+#ifdef HAVE_LIBUNWIND_MIPS_SUPPORT
+#include <libunwind-mips.h>
+#endif
+
+int __get_perf_regnum_for_unw_regnum_mips(int unw_regnum __maybe_unused)
+{
+#ifndef HAVE_LIBUNWIND_MIPS_SUPPORT
+ return -EINVAL;
+#else
+ switch (unw_regnum) {
+ case UNW_MIPS_R1 ... UNW_MIPS_R25:
+ return unw_regnum - UNW_MIPS_R1 + PERF_REG_MIPS_R1;
+ case UNW_MIPS_R28 ... UNW_MIPS_R31:
+ return unw_regnum - UNW_MIPS_R28 + PERF_REG_MIPS_R28;
+ case UNW_MIPS_PC:
+ return PERF_REG_MIPS_PC;
+ default:
+ pr_err("unwind: invalid reg id %d\n", unw_regnum);
+ return -EINVAL;
+ }
+#endif // HAVE_LIBUNWIND_MIPS_SUPPORT
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-ppc32.c b/tools/perf/util/libunwind-arch/libunwind-ppc32.c
new file mode 100644
index 000000000000..edcb0ec95dd7
--- /dev/null
+++ b/tools/perf/util/libunwind-arch/libunwind-ppc32.c
@@ -0,0 +1,31 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+#include "libunwind-arch.h"
+#include "../debug.h"
+#include "../../../arch/powerpc/include/uapi/asm/perf_regs.h"
+#include <linux/compiler.h>
+#include <errno.h>
+
+#ifdef HAVE_LIBUNWIND_PPC32_SUPPORT
+#include <libunwind-ppc32.h>
+#endif
+
+int __get_perf_regnum_for_unw_regnum_ppc32(int unw_regnum __maybe_unused)
+{
+#ifndef HAVE_LIBUNWIND_PPC32_SUPPORT
+ return -EINVAL;
+#else
+ switch (unw_regnum) {
+ case UNW_PPC32_R0 ... UNW_PPC32_31:
+ return unw_regnum - UNW_PPC32_R0 + PERF_REG_POWERPC_R0;
+ case UNW_PPC32_LR:
+ return PERF_REG_POWERPC_LINK;
+ case UNW_PPC32_CTR:
+ return PERF_REG_POWERPC_CTR;
+ case UNW_PPC32_XER:
+ return PERF_REG_POWERPC_XER;
+ default:
+ pr_err("unwind: invalid reg id %d\n", unw_regnum);
+ return -EINVAL;
+ }
+#endif // HAVE_LIBUNWIND_PPC32_SUPPORT
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-ppc64.c b/tools/perf/util/libunwind-arch/libunwind-ppc64.c
new file mode 100644
index 000000000000..9f57a049600b
--- /dev/null
+++ b/tools/perf/util/libunwind-arch/libunwind-ppc64.c
@@ -0,0 +1,33 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+#include "libunwind-arch.h"
+#include "../debug.h"
+#include "../../../arch/powerpc/include/uapi/asm/perf_regs.h"
+#include <linux/compiler.h>
+#include <errno.h>
+
+#ifdef HAVE_LIBUNWIND_PPC64_SUPPORT
+#include <libunwind-ppc64.h>
+#endif
+
+int __get_perf_regnum_for_unw_regnum_ppc64(int unw_regnum __maybe_unused)
+{
+#ifndef HAVE_LIBUNWIND_PPC64_SUPPORT
+ return -EINVAL;
+#else
+ switch (unw_regnum) {
+ case UNW_PPC64_R0 ... UNW_PPC64_31:
+ return unw_regnum - UNW_PPC64_R0 + PERF_REG_POWERPC_R0;
+ case UNW_PPC64_LR:
+ return PERF_REG_POWERPC_LINK;
+ case UNW_PPC64_CTR:
+ return PERF_REG_POWERPC_CTR;
+ case UNW_PPC64_XER:
+ return PERF_REG_POWERPC_XER;
+ case UNW_PPC64_NIP:
+ return PERF_REG_POWERPC_NIP;
+ default:
+ pr_err("unwind: invalid reg id %d\n", unw_regnum);
+ return -EINVAL;
+ }
+#endif // HAVE_LIBUNWIND_PPC64_SUPPORT
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-s390.c b/tools/perf/util/libunwind-arch/libunwind-s390.c
new file mode 100644
index 000000000000..9fcc7885ca55
--- /dev/null
+++ b/tools/perf/util/libunwind-arch/libunwind-s390.c
@@ -0,0 +1,29 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "libunwind-arch.h"
+#include "../debug.h"
+#include "../../../arch/s390/include/uapi/asm/perf_regs.h"
+#include <linux/compiler.h>
+#include <errno.h>
+
+#ifdef HAVE_LIBUNWIND_S390X_SUPPORT
+#include <libunwind-s390x.h>
+#endif
+
+int __get_perf_regnum_for_unw_regnum_s390(int unw_regnum __maybe_unused)
+{
+#ifndef HAVE_LIBUNWIND_S390X_SUPPORT
+ return -EINVAL;
+#else
+ switch (unw_regnum) {
+ case UNW_S390X_R0 ... UNW_S390_R15:
+ return unw_regnum - UNW_S390_R0 + PERF_REG_S390_R0;
+ case UNW_S390X_F0 ... UNW_S390_F15:
+ return unw_regnum - UNW_S390_F0 + PERF_REG_S390_FP0;
+ case UNW_S390X_IP:
+ return PERF_REG_S390_PC;
+ default:
+ pr_err("unwind: invalid reg id %d\n", unw_regnum);
+ return -EINVAL;
+ }
+#endif // HAVE_LIBUNWIND_S390X_SUPPORT
+}
diff --git a/tools/perf/util/libunwind-arch/libunwind-x86_64.c b/tools/perf/util/libunwind-arch/libunwind-x86_64.c
new file mode 100644
index 000000000000..6072e3597e61
--- /dev/null
+++ b/tools/perf/util/libunwind-arch/libunwind-x86_64.c
@@ -0,0 +1,52 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "libunwind-arch.h"
+#include "../debug.h"
+#include "../../../arch/x86/include/uapi/asm/perf_regs.h"
+#include <linux/compiler.h>
+#include <linux/kernel.h>
+#include <errno.h>
+
+#ifdef HAVE_LIBUNWIND_X86_64_SUPPORT
+#include <libunwind-x86_64.h>
+#endif
+
+int __get_perf_regnum_for_unw_regnum_x86_64(int unw_regnum __maybe_unused)
+{
+#ifndef HAVE_LIBUNWIND_X86_64_SUPPORT
+ return -EINVAL;
+#else
+ static const int perf_x86_64_regnums[] = {
+#define REGNUM(reg) [UNW_X86_64_R ## reg] = PERF_REG_X86_ ## reg
+ REGNUM(AX),
+ REGNUM(DX),
+ REGNUM(CX),
+ REGNUM(BX),
+ REGNUM(SI),
+ REGNUM(DI),
+ REGNUM(BP),
+ REGNUM(SP),
+ REGNUM(IP),
+#undef REGNUM
+#define REGNUM(reg) [UNW_X86_64_ ## reg] = PERF_REG_X86_ ## reg
+ REGNUM(R8),
+ REGNUM(R9),
+ REGNUM(R10),
+ REGNUM(R11),
+ REGNUM(R12),
+ REGNUM(R13),
+ REGNUM(R14),
+ REGNUM(R15),
+#undef REGNUM
+ };
+
+ if (unw_regnum == UNW_X86_64_RAX)
+ return PERF_REG_X86_AX;
+
+ if (unw_regnum < 0 || unw_regnum > (int)ARRAY_SIZE(perf_x86_64_regnums) ||
+ perf_x86_64_regnums[unw_regnum] == 0) {
+ pr_err("unwind: invalid reg id %d\n", unw_regnum);
+ return -EINVAL;
+ }
+ return perf_x86_64_regnums[unw_regnum];
+#endif // HAVE_LIBUNWIND_X86_64_SUPPORT
+}
diff --git a/tools/perf/util/libunwind/arm64.c b/tools/perf/util/libunwind/arm64.c
index 37ecef0c53b9..15670a964495 100644
--- a/tools/perf/util/libunwind/arm64.c
+++ b/tools/perf/util/libunwind/arm64.c
@@ -14,11 +14,6 @@
#define REMOTE_UNWIND_LIBUNWIND
-/* Define arch specific functions & regs for libunwind, should be
- * defined before including "unwind.h"
- */
-#define LIBUNWIND__ARCH_REG_ID(regnum) libunwind__arm64_reg_id(regnum)
-
#include "unwind.h"
#include "libunwind-aarch64.h"
#define perf_event_arm_regs perf_event_arm64_regs
diff --git a/tools/perf/util/libunwind/x86_32.c b/tools/perf/util/libunwind/x86_32.c
index 1697dece1b74..1e9fb8bfec44 100644
--- a/tools/perf/util/libunwind/x86_32.c
+++ b/tools/perf/util/libunwind/x86_32.c
@@ -14,20 +14,8 @@
#define REMOTE_UNWIND_LIBUNWIND
-/* Define arch specific functions & regs for libunwind, should be
- * defined before including "unwind.h"
- */
-#define LIBUNWIND__ARCH_REG_ID(regnum) libunwind__x86_reg_id(regnum)
-
#include "unwind.h"
#include "libunwind-x86.h"
-#include <../../../../arch/x86/include/uapi/asm/perf_regs.h>
-
-/* HAVE_ARCH_X86_64_SUPPORT is used in'arch/x86/util/unwind-libunwind.c'
- * for x86_32, we undef it to compile code for x86_32 only.
- */
-#undef HAVE_ARCH_X86_64_SUPPORT
-#include "../../arch/x86/util/unwind-libunwind.c"
/* Explicitly define NO_LIBUNWIND_DEBUG_FRAME, because non-ARM has no
* dwarf_find_debug_frame() function.
diff --git a/tools/perf/util/unwind-libunwind-local.c b/tools/perf/util/unwind-libunwind-local.c
index 5b39ce21e333..3ecdb468b859 100644
--- a/tools/perf/util/unwind-libunwind-local.c
+++ b/tools/perf/util/unwind-libunwind-local.c
@@ -40,6 +40,7 @@
#include "debug.h"
#include "asm/bug.h"
#include "dso.h"
+#include "libunwind-arch/libunwind-arch.h"
extern int
UNW_OBJ(dwarf_search_unwind_table) (unw_addr_space_t as,
@@ -96,6 +97,7 @@ struct unwind_info {
struct perf_sample *sample;
struct machine *machine;
struct thread *thread;
+ uint16_t e_machine;
bool best_effort;
};
@@ -584,9 +586,7 @@ static int access_mem(unw_addr_space_t __maybe_unused as,
}
ret = perf_reg_value(&start, perf_sample__user_regs(ui->sample),
- perf_arch_reg_sp(thread__e_machine(ui->thread,
- ui->machine,
- /*e_flags=*/NULL)));
+ perf_arch_reg_sp(ui->e_machine));
if (ret)
return ret;
@@ -634,7 +634,7 @@ static int access_reg(unw_addr_space_t __maybe_unused as,
return 0;
}
- id = LIBUNWIND__ARCH_REG_ID(regnum);
+ id = get_perf_regnum_for_unw_regnum(ui->e_machine, regnum);
if (id < 0)
return -EINVAL;
@@ -735,7 +735,6 @@ static void _unwind__finish_access(struct maps *maps)
static int get_entries(struct unwind_info *ui, unwind_entry_cb_t cb,
void *arg, int max_stack)
{
- uint16_t e_machine = thread__e_machine(ui->thread, ui->machine, /*e_flags=*/NULL);
u64 val;
unw_word_t ips[max_stack];
unw_addr_space_t addr_space;
@@ -743,7 +742,7 @@ static int get_entries(struct unwind_info *ui, unwind_entry_cb_t cb,
int ret, i = 0;
ret = perf_reg_value(&val, perf_sample__user_regs(ui->sample),
- perf_arch_reg_ip(e_machine));
+ perf_arch_reg_ip(ui->e_machine));
if (ret)
return ret;
@@ -806,6 +805,7 @@ static int _unwind__get_entries(unwind_entry_cb_t cb, void *arg,
.sample = data,
.thread = thread,
.machine = maps__machine(thread__maps(thread)),
+ .e_machine = thread__e_machine(thread, /*machine=*/NULL, /*e_flags=*/NULL),
.best_effort = best_effort
};
diff --git a/tools/perf/util/unwind.h b/tools/perf/util/unwind.h
index ac0776e39f84..f8a8788ac986 100644
--- a/tools/perf/util/unwind.h
+++ b/tools/perf/util/unwind.h
@@ -64,11 +64,6 @@ int libunwind__get_entries(unwind_entry_cb_t cb, void *arg,
struct thread *thread,
struct perf_sample *data, int max_stack,
bool best_effort);
-#ifndef LIBUNWIND__ARCH_REG_ID
-#define LIBUNWIND__ARCH_REG_ID(regnum) libunwind__arch_reg_id(regnum)
-#endif
-
-int LIBUNWIND__ARCH_REG_ID(int regnum);
int unwind__prepare_access(struct maps *maps, struct map *map, bool *initialized);
void unwind__flush_access(struct maps *maps);
void unwind__finish_access(struct maps *maps);
--
2.53.0.1213.gd9a14994de-goog
^ permalink raw reply related
* [PATCH v3 4/8] perf build: Be more programmatic when setting up libunwind variables
From: Ian Rogers @ 2026-04-04 5:40 UTC (permalink / raw)
To: acme
Cc: irogers, 9erthalion6, adrian.hunter, alex, alexander.shishkin,
andrew.jones, aou, atrajeev, blakejones, ctshao, dapeng1.mi,
howardchu95, james.clark, john.g.garry, jolsa, leo.yan,
libunwind-devel, linux-arm-kernel, linux-kernel, linux-perf-users,
linux-riscv, mingo, namhyung, palmer, peterz, pjw, shimin.guo,
tglozar, tmricht, will, yuzhuo
In-Reply-To: <20260404054032.1538095-1-irogers@google.com>
Iterate LIBUNWIND_ARCHS when setting up CONFIG_ and HAVE_ definitions
rather than treating each architecture individually. This sets up the
libunwind build variables and C definitions beyond x86 and
arm/aarch64. The existing naming convention is followed for
compatibility.
Signed-off-by: Ian Rogers <irogers@google.com>
---
tools/perf/Makefile.config | 215 +++++++++++++++----------------------
1 file changed, 89 insertions(+), 126 deletions(-)
diff --git a/tools/perf/Makefile.config b/tools/perf/Makefile.config
index 333ddd0e4bd8..9d31d0f6f52a 100644
--- a/tools/perf/Makefile.config
+++ b/tools/perf/Makefile.config
@@ -65,95 +65,83 @@ $(call detected_var,SRCARCH)
CFLAGS += -I$(OUTPUT)arch/$(SRCARCH)/include/generated
-# Additional ARCH settings for ppc
-ifeq ($(SRCARCH),powerpc)
- ifndef NO_LIBUNWIND
- LIBUNWIND_LIBS := -lunwind -lunwind-ppc64
- endif
-endif
-
# Additional ARCH settings for x86
ifeq ($(SRCARCH),x86)
$(call detected,CONFIG_X86)
ifeq (${IS_64_BIT}, 1)
CFLAGS += -DHAVE_ARCH_X86_64_SUPPORT
ARCH_INCLUDE = ../../arch/x86/lib/memcpy_64.S ../../arch/x86/lib/memset_64.S
- ifndef NO_LIBUNWIND
- LIBUNWIND_LIBS = -lunwind-x86_64 -lunwind -llzma
- endif
$(call detected,CONFIG_X86_64)
- else
- ifndef NO_LIBUNWIND
- LIBUNWIND_LIBS = -lunwind-x86 -llzma -lunwind
- endif
endif
endif
-ifeq ($(SRCARCH),arm)
- ifndef NO_LIBUNWIND
- LIBUNWIND_LIBS = -lunwind -lunwind-arm
- endif
+ifeq ($(ARCH),s390)
+ CFLAGS += -fPIC
endif
-ifeq ($(SRCARCH),arm64)
- ifndef NO_LIBUNWIND
- LIBUNWIND_LIBS = -lunwind -lunwind-aarch64
- endif
+ifneq ($(LIBUNWIND),1)
+ NO_LIBUNWIND := 1
endif
-ifeq ($(SRCARCH),loongarch)
- ifndef NO_LIBUNWIND
+ifndef NO_LIBUNWIND
+ ifeq ($(SRCARCH),arm)
+ LIBUNWIND_LIBS = -lunwind -lunwind-arm
+ endif
+ ifeq ($(SRCARCH),arm64)
+ LIBUNWIND_LIBS = -lunwind -lunwind-aarch64
+ endif
+ ifeq ($(SRCARCH),loongarch)
LIBUNWIND_LIBS = -lunwind -lunwind-loongarch64
endif
-endif
-
-ifeq ($(ARCH),s390)
- CFLAGS += -fPIC
-endif
-
-ifeq ($(ARCH),mips)
- ifndef NO_LIBUNWIND
+ ifeq ($(ARCH),mips)
LIBUNWIND_LIBS = -lunwind -lunwind-mips
endif
+ ifeq ($(SRCARCH),powerpc)
+ LIBUNWIND_LIBS := -lunwind -lunwind-ppc64
+ endif
+ ifeq ($(SRCARCH),riscv)
+ LIBUNWIND_LIBS := -lunwind -lunwind-riscv
+ endif
+ ifeq ($(SRCARCH),s390)
+ LIBUNWIND_LIBS := -lunwind -lunwind-s390x
+ endif
+ ifeq ($(SRCARCH),x86)
+ ifeq (${IS_64_BIT}, 1)
+ LIBUNWIND_LIBS = -lunwind-x86_64 -lunwind -llzma
+ else
+ LIBUNWIND_LIBS = -lunwind-x86 -lunwind -llzma
+ endif
+ endif
+ ifeq ($(LIBUNWIND_LIBS),)
+ NO_LIBUNWIND := 1
+ endif
endif
-ifneq ($(LIBUNWIND),1)
- NO_LIBUNWIND := 1
-endif
-
-ifeq ($(LIBUNWIND_LIBS),)
- NO_LIBUNWIND := 1
-endif
#
# For linking with debug library, run like:
#
# make DEBUG=1 LIBUNWIND_DIR=/opt/libunwind/
#
-
-libunwind_arch_set_flags = $(eval $(libunwind_arch_set_flags_code))
-define libunwind_arch_set_flags_code
- FEATURE_CHECK_CFLAGS-libunwind-$(1) = -I$(LIBUNWIND_DIR)/include
- FEATURE_CHECK_LDFLAGS-libunwind-$(1) = -L$(LIBUNWIND_DIR)/lib
-endef
-
-ifdef LIBUNWIND_DIR
- LIBUNWIND_CFLAGS = -I$(LIBUNWIND_DIR)/include
- LIBUNWIND_LDFLAGS = -L$(LIBUNWIND_DIR)/lib
- LIBUNWIND_ARCHS = x86 x86_64 arm aarch64 debug-frame-arm debug-frame-aarch64 loongarch
- $(foreach libunwind_arch,$(LIBUNWIND_ARCHS),$(call libunwind_arch_set_flags,$(libunwind_arch)))
-endif
+LIBUNWIND_ARCHS:=aarch64 arm loongarch64 mips ppc32 ppc64 riscv s390x x86 x86_64
ifndef NO_LIBUNWIND
- # Set per-feature check compilation flags
FEATURE_CHECK_CFLAGS-libunwind = $(LIBUNWIND_CFLAGS)
FEATURE_CHECK_LDFLAGS-libunwind = $(LIBUNWIND_LDFLAGS) $(LIBUNWIND_LIBS)
FEATURE_CHECK_CFLAGS-libunwind-debug-frame = $(LIBUNWIND_CFLAGS)
FEATURE_CHECK_LDFLAGS-libunwind-debug-frame = $(LIBUNWIND_LDFLAGS) $(LIBUNWIND_LIBS)
-
- FEATURE_CHECK_LDFLAGS-libunwind-arm += -lunwind -lunwind-arm
- FEATURE_CHECK_LDFLAGS-libunwind-aarch64 += -lunwind -lunwind-aarch64
- FEATURE_CHECK_LDFLAGS-libunwind-x86 += -lunwind -llzma -lunwind-x86
- FEATURE_CHECK_LDFLAGS-libunwind-x86_64 += -lunwind -llzma -lunwind-x86_64
+
+ ifdef LIBUNWIND_DIR
+ LIBUNWIND_CFLAGS = -I$(LIBUNWIND_DIR)/include
+ LIBUNWIND_LDFLAGS = -L$(LIBUNWIND_DIR)/lib
+
+ define libunwind_arch_set_flags
+ FEATURE_CHECK_CFLAGS-libunwind-$(1) = -I$(LIBUNWIND_DIR)/include
+ FEATURE_CHECK_LDFLAGS-libunwind-$(1) = -L$(LIBUNWIND_DIR)/lib
+ endef
+ $(foreach arch,$(LIBUNWIND_ARCHS), \
+ $(eval $(call libunwind_arch_set_flags,$(arch))) \
+ )
+ endif
endif
ifdef CSINCLUDES
@@ -638,49 +626,6 @@ ifeq ($(SRCARCH),powerpc)
endif
endif
-ifndef NO_LIBUNWIND
- have_libunwind :=
-
- $(call feature_check,libunwind)
-
- $(call feature_check,libunwind-x86)
- ifeq ($(feature-libunwind-x86), 1)
- $(call detected,CONFIG_LIBUNWIND_X86)
- CFLAGS += -DHAVE_LIBUNWIND_X86_SUPPORT
- LDFLAGS += -lunwind-x86
- EXTLIBS_LIBUNWIND += -lunwind-x86
- have_libunwind = 1
- endif
-
- $(call feature_check,libunwind-aarch64)
- ifeq ($(feature-libunwind-aarch64), 1)
- $(call detected,CONFIG_LIBUNWIND_AARCH64)
- CFLAGS += -DHAVE_LIBUNWIND_AARCH64_SUPPORT
- LDFLAGS += -lunwind-aarch64
- EXTLIBS_LIBUNWIND += -lunwind-aarch64
- have_libunwind = 1
- $(call feature_check,libunwind-debug-frame-aarch64)
- ifneq ($(feature-libunwind-debug-frame-aarch64), 1)
- $(warning No debug_frame support found in libunwind-aarch64)
- CFLAGS += -DNO_LIBUNWIND_DEBUG_FRAME_AARCH64
- endif
- endif
-
- ifneq ($(feature-libunwind), 1)
- $(warning No libunwind found. Please install libunwind-dev[el] >= 1.1 and/or set LIBUNWIND_DIR and set LIBUNWIND=1 in the make command line as it is opt-in now)
- NO_LOCAL_LIBUNWIND := 1
- else
- have_libunwind := 1
- $(call detected,CONFIG_LOCAL_LIBUNWIND)
- endif
-
- ifneq ($(have_libunwind), 1)
- NO_LIBUNWIND := 1
- endif
-else
- NO_LOCAL_LIBUNWIND := 1
-endif
-
ifndef NO_LIBBPF
ifneq ($(feature-bpf), 1)
$(warning BPF API too old. Please install recent kernel headers. BPF support in 'perf record' is disabled.)
@@ -739,6 +684,49 @@ ifndef GEN_VMLINUX_H
VMLINUX_H=$(src-perf)/util/bpf_skel/vmlinux/vmlinux.h
endif
+ifndef NO_LIBUNWIND
+ have_libunwind :=
+
+ $(call feature_check,libunwind)
+ ifneq ($(feature-libunwind), 1)
+ $(warning No libunwind found. Please install libunwind-dev[el] >= 1.1 and/or set LIBUNWIND_DIR and set LIBUNWIND=1 in the make command line as it is opt-in now)
+ NO_LOCAL_LIBUNWIND := 1
+ else
+ have_libunwind := 1
+ $(call detected,CONFIG_LOCAL_LIBUNWIND)
+ CFLAGS += -DHAVE_LIBUNWIND_SUPPORT
+ CFLAGS += $(LIBUNWIND_CFLAGS)
+ LDFLAGS += $(LIBUNWIND_LDFLAGS)
+ EXTLIBS += $(EXTLIBS_LIBUNWIND)
+ $(call feature_check,libunwind-debug-frame)
+ ifneq ($(feature-libunwind-debug-frame), 1)
+ CFLAGS += -DNO_LIBUNWIND_DEBUG_FRAME
+ endif
+ endif
+
+ define PROCESS_REMOTE_LIBUNWIND_ARCH
+ $(call feature_check,libunwind-$(1))
+
+ ifeq ($$(feature-libunwind-$(1)), 1)
+ upper_arch := $$(shell echo $(1) | tr '[:lower:]' '[:upper:]')
+ $$(call detected,CONFIG_LIBUNWIND_$$(upper_arch))
+
+ CFLAGS += -DHAVE_LIBUNWIND_$$(upper_arch)_SUPPORT
+ LDFLAGS += -lunwind-$(1)
+ EXTLIBS_LIBUNWIND += -lunwind-$(1)
+ have_libunwind := 1
+
+ $$(call feature_check,libunwind-debug-frame-$(1))
+ ifneq ($$(feature-libunwind-debug-frame-$(1)), 1)
+ CFLAGS += -DNO_LIBUNWIND_DEBUG_FRAME_$$(upper_arch)
+ endif
+ endif
+ endef
+ $(foreach arch,$(LIBUNWIND_ARCHS), \
+ $(eval $(call PROCESS_REMOTE_LIBUNWIND_ARCH,$(arch))) \
+ )
+endif
+
dwarf-post-unwind := 1
dwarf-post-unwind-text := BUG
@@ -761,31 +749,6 @@ ifeq ($(dwarf-post-unwind),1)
$(call detected,CONFIG_DWARF_UNWIND)
endif
-ifndef NO_LIBUNWIND
- ifndef NO_LOCAL_LIBUNWIND
- ifeq ($(SRCARCH),$(filter $(SRCARCH),arm arm64))
- $(call feature_check,libunwind-debug-frame)
- ifneq ($(feature-libunwind-debug-frame), 1)
- $(warning No debug_frame support found in libunwind)
- CFLAGS += -DNO_LIBUNWIND_DEBUG_FRAME
- endif
- else
- # non-ARM has no dwarf_find_debug_frame() function:
- CFLAGS += -DNO_LIBUNWIND_DEBUG_FRAME
- endif
- EXTLIBS += $(LIBUNWIND_LIBS)
- LDFLAGS += $(LIBUNWIND_LIBS)
- endif
- ifeq ($(findstring -static,${LDFLAGS}),-static)
- # gcc -static links libgcc_eh which contans piece of libunwind
- LIBUNWIND_LDFLAGS += -Wl,--allow-multiple-definition
- endif
- CFLAGS += -DHAVE_LIBUNWIND_SUPPORT
- CFLAGS += $(LIBUNWIND_CFLAGS)
- LDFLAGS += $(LIBUNWIND_LDFLAGS)
- EXTLIBS += $(EXTLIBS_LIBUNWIND)
-endif
-
ifneq ($(NO_LIBTRACEEVENT),1)
$(call detected,CONFIG_TRACE)
endif
--
2.53.0.1213.gd9a14994de-goog
^ permalink raw reply related
* [PATCH v3 3/8] tools build: Deduplicate test-libunwind for different architectures
From: Ian Rogers @ 2026-04-04 5:40 UTC (permalink / raw)
To: acme
Cc: irogers, 9erthalion6, adrian.hunter, alex, alexander.shishkin,
andrew.jones, aou, atrajeev, blakejones, ctshao, dapeng1.mi,
howardchu95, james.clark, john.g.garry, jolsa, leo.yan,
libunwind-devel, linux-arm-kernel, linux-kernel, linux-perf-users,
linux-riscv, mingo, namhyung, palmer, peterz, pjw, shimin.guo,
tglozar, tmricht, will, yuzhuo
In-Reply-To: <20260404054032.1538095-1-irogers@google.com>
The separate test files only exist to pass a different #include,
instead have a single source file and pass -include to $(CC) to
include the relevant header file for the architecture being
tested. Generate the rules using a foreach loop. Include tests for all
current libunwind architectures.
Signed-off-by: Ian Rogers <irogers@google.com>
---
tools/build/feature/Makefile | 38 +++++++++----------
tools/build/feature/test-libunwind-aarch64.c | 27 -------------
tools/build/feature/test-libunwind-arm.c | 28 --------------
.../test-libunwind-debug-frame-aarch64.c | 17 ---------
.../feature/test-libunwind-debug-frame-arm.c | 17 ---------
.../feature/test-libunwind-debug-frame.c | 1 -
tools/build/feature/test-libunwind-x86.c | 28 --------------
tools/build/feature/test-libunwind-x86_64.c | 28 --------------
tools/build/feature/test-libunwind.c | 1 -
9 files changed, 19 insertions(+), 166 deletions(-)
delete mode 100644 tools/build/feature/test-libunwind-aarch64.c
delete mode 100644 tools/build/feature/test-libunwind-arm.c
delete mode 100644 tools/build/feature/test-libunwind-debug-frame-aarch64.c
delete mode 100644 tools/build/feature/test-libunwind-debug-frame-arm.c
delete mode 100644 tools/build/feature/test-libunwind-x86.c
delete mode 100644 tools/build/feature/test-libunwind-x86_64.c
diff --git a/tools/build/feature/Makefile b/tools/build/feature/Makefile
index f163a245837a..bda8ef868e44 100644
--- a/tools/build/feature/Makefile
+++ b/tools/build/feature/Makefile
@@ -210,27 +210,27 @@ $(OUTPUT)test-numa_num_possible_cpus.bin:
$(BUILD) -lnuma
$(OUTPUT)test-libunwind.bin:
- $(BUILD) -lelf -llzma
+ $(BUILD) -include libunwind.h -lelf -llzma -lunwind
$(OUTPUT)test-libunwind-debug-frame.bin:
- $(BUILD) -lelf -llzma
-$(OUTPUT)test-libunwind-x86.bin:
- $(BUILD) -lelf -llzma -lunwind-x86
-
-$(OUTPUT)test-libunwind-x86_64.bin:
- $(BUILD) -lelf -llzma -lunwind-x86_64
-
-$(OUTPUT)test-libunwind-arm.bin:
- $(BUILD) -lelf -llzma -lunwind-arm
-
-$(OUTPUT)test-libunwind-aarch64.bin:
- $(BUILD) -lelf -llzma -lunwind-aarch64
-
-$(OUTPUT)test-libunwind-debug-frame-arm.bin:
- $(BUILD) -lelf -llzma -lunwind-arm
-
-$(OUTPUT)test-libunwind-debug-frame-aarch64.bin:
- $(BUILD) -lelf -llzma -lunwind-aarch64
+ $(BUILD) -include libunwind.h -lelf -llzma -lunwind
+
+LIBUNWIND_ARCHS:=aarch64 arm loongarch64 mips ppc32 ppc64 riscv s390x x86 x86_64
+define LIBUNWIND_RULE
+$$(OUTPUT)test-libunwind-$(1).bin:
+ $$(CC) $$(CFLAGS) -MD -Wall -Werror -include libunwind-$(1).h -o $$@ \
+ test-libunwind.c $$(LDFLAGS) -lelf -llzma -lunwind-$(1) \
+ > $$(@:.bin=.make.output) 2>&1
+
+$$(OUTPUT)test-libunwind-debug-frame-$(1).bin:
+ $$(CC) $$(CFLAGS) -MD -Wall -Werror -include libunwind-$(1).h -o $$@ \
+ test-libunwind-debug-frame.c $$(LDFLAGS) -lelf -llzma -lunwind-$(1) \
+ > $$(@:.bin=.make.output) 2>&1
+
+endef
+$(foreach arch,$(LIBUNWIND_ARCHS), \
+ $(eval $(call LIBUNWIND_RULE,$(arch))) \
+)
$(OUTPUT)test-libslang.bin:
$(BUILD) -lslang
diff --git a/tools/build/feature/test-libunwind-aarch64.c b/tools/build/feature/test-libunwind-aarch64.c
deleted file mode 100644
index 323803f49212..000000000000
--- a/tools/build/feature/test-libunwind-aarch64.c
+++ /dev/null
@@ -1,27 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-#include <libunwind-aarch64.h>
-#include <stdlib.h>
-
-extern int UNW_OBJ(dwarf_search_unwind_table) (unw_addr_space_t as,
- unw_word_t ip,
- unw_dyn_info_t *di,
- unw_proc_info_t *pi,
- int need_unwind_info, void *arg);
-
-#define dwarf_search_unwind_table UNW_OBJ(dwarf_search_unwind_table)
-
-static unw_accessors_t accessors;
-
-int main(void)
-{
- unw_addr_space_t addr_space;
-
- addr_space = unw_create_addr_space(&accessors, 0);
- if (addr_space)
- return 0;
-
- unw_init_remote(NULL, addr_space, NULL);
- dwarf_search_unwind_table(addr_space, 0, NULL, NULL, 0, NULL);
-
- return 0;
-}
diff --git a/tools/build/feature/test-libunwind-arm.c b/tools/build/feature/test-libunwind-arm.c
deleted file mode 100644
index cb378b7d6866..000000000000
--- a/tools/build/feature/test-libunwind-arm.c
+++ /dev/null
@@ -1,28 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-#include <libunwind-arm.h>
-#include <stdlib.h>
-
-extern int UNW_OBJ(dwarf_search_unwind_table) (unw_addr_space_t as,
- unw_word_t ip,
- unw_dyn_info_t *di,
- unw_proc_info_t *pi,
- int need_unwind_info, void *arg);
-
-
-#define dwarf_search_unwind_table UNW_OBJ(dwarf_search_unwind_table)
-
-static unw_accessors_t accessors;
-
-int main(void)
-{
- unw_addr_space_t addr_space;
-
- addr_space = unw_create_addr_space(&accessors, 0);
- if (addr_space)
- return 0;
-
- unw_init_remote(NULL, addr_space, NULL);
- dwarf_search_unwind_table(addr_space, 0, NULL, NULL, 0, NULL);
-
- return 0;
-}
diff --git a/tools/build/feature/test-libunwind-debug-frame-aarch64.c b/tools/build/feature/test-libunwind-debug-frame-aarch64.c
deleted file mode 100644
index 36d6646c185e..000000000000
--- a/tools/build/feature/test-libunwind-debug-frame-aarch64.c
+++ /dev/null
@@ -1,17 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-#include <libunwind-aarch64.h>
-#include <stdlib.h>
-
-extern int
-UNW_OBJ(dwarf_find_debug_frame) (int found, unw_dyn_info_t *di_debug,
- unw_word_t ip, unw_word_t segbase,
- const char *obj_name, unw_word_t start,
- unw_word_t end);
-
-#define dwarf_find_debug_frame UNW_OBJ(dwarf_find_debug_frame)
-
-int main(void)
-{
- dwarf_find_debug_frame(0, NULL, 0, 0, NULL, 0, 0);
- return 0;
-}
diff --git a/tools/build/feature/test-libunwind-debug-frame-arm.c b/tools/build/feature/test-libunwind-debug-frame-arm.c
deleted file mode 100644
index 8696e48e1268..000000000000
--- a/tools/build/feature/test-libunwind-debug-frame-arm.c
+++ /dev/null
@@ -1,17 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-#include <libunwind-arm.h>
-#include <stdlib.h>
-
-extern int
-UNW_OBJ(dwarf_find_debug_frame) (int found, unw_dyn_info_t *di_debug,
- unw_word_t ip, unw_word_t segbase,
- const char *obj_name, unw_word_t start,
- unw_word_t end);
-
-#define dwarf_find_debug_frame UNW_OBJ(dwarf_find_debug_frame)
-
-int main(void)
-{
- dwarf_find_debug_frame(0, NULL, 0, 0, NULL, 0, 0);
- return 0;
-}
diff --git a/tools/build/feature/test-libunwind-debug-frame.c b/tools/build/feature/test-libunwind-debug-frame.c
index efb55cdd8d01..4c57e37004b3 100644
--- a/tools/build/feature/test-libunwind-debug-frame.c
+++ b/tools/build/feature/test-libunwind-debug-frame.c
@@ -1,5 +1,4 @@
// SPDX-License-Identifier: GPL-2.0
-#include <libunwind.h>
#include <stdlib.h>
extern int
diff --git a/tools/build/feature/test-libunwind-x86.c b/tools/build/feature/test-libunwind-x86.c
deleted file mode 100644
index e5e0f6c89637..000000000000
--- a/tools/build/feature/test-libunwind-x86.c
+++ /dev/null
@@ -1,28 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-#include <libunwind-x86.h>
-#include <stdlib.h>
-
-extern int UNW_OBJ(dwarf_search_unwind_table) (unw_addr_space_t as,
- unw_word_t ip,
- unw_dyn_info_t *di,
- unw_proc_info_t *pi,
- int need_unwind_info, void *arg);
-
-
-#define dwarf_search_unwind_table UNW_OBJ(dwarf_search_unwind_table)
-
-static unw_accessors_t accessors;
-
-int main(void)
-{
- unw_addr_space_t addr_space;
-
- addr_space = unw_create_addr_space(&accessors, 0);
- if (addr_space)
- return 0;
-
- unw_init_remote(NULL, addr_space, NULL);
- dwarf_search_unwind_table(addr_space, 0, NULL, NULL, 0, NULL);
-
- return 0;
-}
diff --git a/tools/build/feature/test-libunwind-x86_64.c b/tools/build/feature/test-libunwind-x86_64.c
deleted file mode 100644
index 62ae4db597dc..000000000000
--- a/tools/build/feature/test-libunwind-x86_64.c
+++ /dev/null
@@ -1,28 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-#include <libunwind-x86_64.h>
-#include <stdlib.h>
-
-extern int UNW_OBJ(dwarf_search_unwind_table) (unw_addr_space_t as,
- unw_word_t ip,
- unw_dyn_info_t *di,
- unw_proc_info_t *pi,
- int need_unwind_info, void *arg);
-
-
-#define dwarf_search_unwind_table UNW_OBJ(dwarf_search_unwind_table)
-
-static unw_accessors_t accessors;
-
-int main(void)
-{
- unw_addr_space_t addr_space;
-
- addr_space = unw_create_addr_space(&accessors, 0);
- if (addr_space)
- return 0;
-
- unw_init_remote(NULL, addr_space, NULL);
- dwarf_search_unwind_table(addr_space, 0, NULL, NULL, 0, NULL);
-
- return 0;
-}
diff --git a/tools/build/feature/test-libunwind.c b/tools/build/feature/test-libunwind.c
index 53fd26614ff0..5af5dc3a73d4 100644
--- a/tools/build/feature/test-libunwind.c
+++ b/tools/build/feature/test-libunwind.c
@@ -1,5 +1,4 @@
// SPDX-License-Identifier: GPL-2.0
-#include <libunwind.h>
#include <stdlib.h>
extern int UNW_OBJ(dwarf_search_unwind_table) (unw_addr_space_t as,
--
2.53.0.1213.gd9a14994de-goog
^ permalink raw reply related
* [PATCH v3 0/8] perf libunwind multiple remote support
From: Ian Rogers @ 2026-04-04 5:40 UTC (permalink / raw)
To: acme
Cc: irogers, 9erthalion6, adrian.hunter, alex, alexander.shishkin,
andrew.jones, aou, atrajeev, blakejones, ctshao, dapeng1.mi,
howardchu95, james.clark, john.g.garry, jolsa, leo.yan,
libunwind-devel, linux-arm-kernel, linux-kernel, linux-perf-users,
linux-riscv, mingo, namhyung, palmer, peterz, pjw, shimin.guo,
tglozar, tmricht, will, yuzhuo
In-Reply-To: <20260305221927.3237145-1-irogers@google.com>
Fix the libunwind build for when libdw and libunwind are feature
detected, currently failing with a duplicate symbol.
Refactor the libunwind support so that whenever a remote target is
available, perf functions using the ELF machine can use that remote
target regardless of what the host/local machine is. Migrate existing
libunwind supported architectures like powerpc, arm64 and loongarch so
that they can work in a cross-architecture way. Add support for
RISC-V. Make the code more regular in function names, etc. and avoid
including a C-file. This increases the lines of code. It is similar in
style to the unwind-libdw implementation. It is hoped that the more
uniform nature of the code with help with refactoring the perf
registers for SIMD/APX support.
Aside from local host testing these patches are under tested, in part
as I'm failing to see how to build libunwind with support for multiple
remote targets. Please could I get help in testing.
v3: Minor whitespace clean up and warn when a dynamic choice of libdw
or libunwind is selected for unwinding and support is missing (Arnaldo).
v2: Move two fixes patches to position 1 and 2 in the series. Fix
struct naming inconsistency, Andrew Jones
<andrew.jones@oss.qualcomm.com>. Fix other inconsistencies and
potential non-x86 build issues.
https://lore.kernel.org/lkml/20260305221927.3237145-1-irogers@google.com/
v1: https://lore.kernel.org/lkml/20260224142938.26088-1-irogers@google.com/
Ian Rogers (8):
perf unwind: Refactor get_entries to allow dynamic libdw/libunwind
selection
perf build loongarch: Remove reference to missing file
tools build: Deduplicate test-libunwind for different architectures
perf build: Be more programmatic when setting up libunwind variables
perf unwind-libunwind: Make libunwind register reading cross platform
perf unwind-libunwind: Move flush/finish access out of local
perf unwind-libunwind: Remove libunwind-local
perf unwind-libunwind: Add RISC-V libunwind support
tools/build/feature/Makefile | 38 +-
tools/build/feature/test-libunwind-aarch64.c | 27 -
tools/build/feature/test-libunwind-arm.c | 28 -
.../test-libunwind-debug-frame-aarch64.c | 17 -
.../feature/test-libunwind-debug-frame-arm.c | 17 -
.../feature/test-libunwind-debug-frame.c | 1 -
tools/build/feature/test-libunwind-x86.c | 28 -
tools/build/feature/test-libunwind-x86_64.c | 28 -
tools/build/feature/test-libunwind.c | 1 -
tools/perf/Makefile.config | 215 ++---
tools/perf/arch/arm/util/Build | 2 -
tools/perf/arch/arm/util/unwind-libunwind.c | 50 --
tools/perf/arch/arm64/util/Build | 1 -
tools/perf/arch/arm64/util/unwind-libunwind.c | 17 -
tools/perf/arch/loongarch/util/Build | 3 -
.../arch/loongarch/util/unwind-libunwind.c | 82 --
tools/perf/arch/mips/Build | 1 -
tools/perf/arch/mips/util/Build | 1 -
tools/perf/arch/mips/util/unwind-libunwind.c | 22 -
tools/perf/arch/powerpc/util/Build | 1 -
.../perf/arch/powerpc/util/unwind-libunwind.c | 92 --
tools/perf/arch/x86/util/Build | 3 -
tools/perf/arch/x86/util/unwind-libunwind.c | 115 ---
tools/perf/builtin-inject.c | 4 +
tools/perf/builtin-report.c | 4 +
tools/perf/builtin-script.c | 4 +
tools/perf/util/Build | 5 +-
tools/perf/util/libunwind-arch/Build | 11 +
.../perf/util/libunwind-arch/libunwind-arch.c | 319 +++++++
.../perf/util/libunwind-arch/libunwind-arch.h | 296 +++++++
.../perf/util/libunwind-arch/libunwind-arm.c | 290 ++++++
.../util/libunwind-arch/libunwind-arm64.c | 289 ++++++
.../perf/util/libunwind-arch/libunwind-i386.c | 312 +++++++
.../util/libunwind-arch/libunwind-loongarch.c | 297 +++++++
.../perf/util/libunwind-arch/libunwind-mips.c | 299 +++++++
.../util/libunwind-arch/libunwind-ppc32.c | 301 +++++++
.../util/libunwind-arch/libunwind-ppc64.c | 303 +++++++
.../util/libunwind-arch/libunwind-riscv.c | 297 +++++++
.../perf/util/libunwind-arch/libunwind-s390.c | 299 +++++++
.../util/libunwind-arch/libunwind-x86_64.c | 320 +++++++
tools/perf/util/libunwind/arm64.c | 40 -
tools/perf/util/libunwind/x86_32.c | 41 -
tools/perf/util/maps.c | 29 +-
tools/perf/util/maps.h | 4 +-
tools/perf/util/symbol_conf.h | 10 +
tools/perf/util/thread.c | 29 +-
tools/perf/util/unwind-libdw.c | 2 +-
tools/perf/util/unwind-libunwind-local.c | 832 ------------------
tools/perf/util/unwind-libunwind.c | 679 ++++++++++++--
tools/perf/util/unwind.c | 98 +++
tools/perf/util/unwind.h | 77 +-
51 files changed, 4549 insertions(+), 1732 deletions(-)
delete mode 100644 tools/build/feature/test-libunwind-aarch64.c
delete mode 100644 tools/build/feature/test-libunwind-arm.c
delete mode 100644 tools/build/feature/test-libunwind-debug-frame-aarch64.c
delete mode 100644 tools/build/feature/test-libunwind-debug-frame-arm.c
delete mode 100644 tools/build/feature/test-libunwind-x86.c
delete mode 100644 tools/build/feature/test-libunwind-x86_64.c
delete mode 100644 tools/perf/arch/arm/util/unwind-libunwind.c
delete mode 100644 tools/perf/arch/arm64/util/unwind-libunwind.c
delete mode 100644 tools/perf/arch/loongarch/util/unwind-libunwind.c
delete mode 100644 tools/perf/arch/mips/Build
delete mode 100644 tools/perf/arch/mips/util/Build
delete mode 100644 tools/perf/arch/mips/util/unwind-libunwind.c
delete mode 100644 tools/perf/arch/powerpc/util/unwind-libunwind.c
delete mode 100644 tools/perf/arch/x86/util/unwind-libunwind.c
create mode 100644 tools/perf/util/libunwind-arch/Build
create mode 100644 tools/perf/util/libunwind-arch/libunwind-arch.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-arch.h
create mode 100644 tools/perf/util/libunwind-arch/libunwind-arm.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-arm64.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-i386.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-loongarch.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-mips.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-ppc32.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-ppc64.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-riscv.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-s390.c
create mode 100644 tools/perf/util/libunwind-arch/libunwind-x86_64.c
delete mode 100644 tools/perf/util/libunwind/arm64.c
delete mode 100644 tools/perf/util/libunwind/x86_32.c
delete mode 100644 tools/perf/util/unwind-libunwind-local.c
create mode 100644 tools/perf/util/unwind.c
--
2.53.0.1213.gd9a14994de-goog
^ permalink raw reply
* Re: [PATCH v3] PCI: imx6: Don't remove MSI capability for i.MX7D/i.MX8M
From: Manivannan Sadhasivam @ 2026-04-04 2:33 UTC (permalink / raw)
To: Bjorn Helgaas
Cc: Richard Zhu, frank.li, l.stach, lpieralisi, kwilczynski, robh,
bhelgaas, s.hauer, kernel, festevam, linux-pci, linux-arm-kernel,
imx, linux-kernel, stable
In-Reply-To: <20260403215656.GA360979@bhelgaas>
On Fri, Apr 03, 2026 at 04:56:56PM -0500, Bjorn Helgaas wrote:
> On Tue, Mar 31, 2026 at 04:52:52PM +0800, Richard Zhu wrote:
> > The MSI trigger mechanism for endpoint devices connected to i.MX7D,
> > i.MX8MM, and i.MX8MQ PCIe root complex ports depends on the MSI
> > capability register settings in the root complex. Removing the MSI
> > capability breaks MSI functionality for these endpoints.
> >
> > Add keep_rp_msi_en flag to indicate platforms (i.MX7D, i.MX8MM, i.MX8MQ)
> > that should preserve the MSI capability during initialization.
>
> I guess Mani added this to the commit log:
>
> Note that by keeping Root Port MSI capability, Root Port MSIs such as AER,
> PME and others won't be received by default. So users need to use
> workarounds such as passing 'pcie_pme=nomsi' cmdline param.
>
Yes, I did.
> Why can't we fix this automatically? I hate it when users are
> expected to use command line parameters to work around issues.
>
> Obviously we know at probe-time, before any PCI devices are
> enumerated, so it seems like we should be able to do the equivalent of
> "pcie_pme=nomsi" even without the parameter.
>
That's why we came up with the idea of disabling the MSI/MSI-X capabilities of
the Root Port so that the code falls back to INTx. But the issue is that,
disabling that capability or MSI Enable bit (mayebe?) prevents endpoint MSIs
from being received. We don't know exactly what's going on inside the hardware
though. But is broken at its best. Fortunately, this behavior only applies to a
subset of the NXP SoCs.
- Mani
> > Cc: stable@vger.kernel.org
> > Fixes: f5cd8a929c825 ("PCI: dwc: Remove MSI/MSIX capability for Root Port if iMSI-RX is used as MSI controller")
> > Suggested-by: Manivannan Sadhasivam <mani@kernel.org>
> > Signed-off-by: Richard Zhu <hongxing.zhu@nxp.com>
> > ---
> > v3 changes:
> > Use a flag 'dw_pcie_rp::keep_rp_msi_en' to identify SoCs that require MSI
> > capability preservation, and skip the capability removal in
> > pcie-designware-host.c accordingly.
> >
> > v2 changes:
> > CC stable tree.
> > ---
> > drivers/pci/controller/dwc/pci-imx6.c | 7 +++++++
> > drivers/pci/controller/dwc/pcie-designware-host.c | 2 +-
> > drivers/pci/controller/dwc/pcie-designware.h | 1 +
> > 3 files changed, 9 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/pci/controller/dwc/pci-imx6.c b/drivers/pci/controller/dwc/pci-imx6.c
> > index 20dafd2710a3..fde173770933 100644
> > --- a/drivers/pci/controller/dwc/pci-imx6.c
> > +++ b/drivers/pci/controller/dwc/pci-imx6.c
> > @@ -117,6 +117,8 @@ enum imx_pcie_variants {
> > #define IMX_PCIE_FLAG_HAS_LUT BIT(10)
> > #define IMX_PCIE_FLAG_8GT_ECN_ERR051586 BIT(11)
> > #define IMX_PCIE_FLAG_SKIP_L23_READY BIT(12)
> > +/* Preserve MSI capability for platforms that require it */
> > +#define IMX_PCIE_FLAG_KEEP_MSI_CAP BIT(13)
> >
> > #define imx_check_flag(pci, val) (pci->drvdata->flags & val)
> >
> > @@ -1820,6 +1822,8 @@ static int imx_pcie_probe(struct platform_device *pdev)
> > } else {
> > if (imx_check_flag(imx_pcie, IMX_PCIE_FLAG_SKIP_L23_READY))
> > pci->pp.skip_l23_ready = true;
> > + if (imx_check_flag(imx_pcie, IMX_PCIE_FLAG_KEEP_MSI_CAP))
> > + pci->pp.keep_rp_msi_en = true;
> > pci->pp.use_atu_msg = true;
> > ret = dw_pcie_host_init(&pci->pp);
> > if (ret < 0)
> > @@ -1897,6 +1901,7 @@ static const struct imx_pcie_drvdata drvdata[] = {
> > [IMX7D] = {
> > .variant = IMX7D,
> > .flags = IMX_PCIE_FLAG_SUPPORTS_SUSPEND |
> > + IMX_PCIE_FLAG_KEEP_MSI_CAP |
> > IMX_PCIE_FLAG_HAS_APP_RESET |
> > IMX_PCIE_FLAG_SKIP_L23_READY |
> > IMX_PCIE_FLAG_HAS_PHY_RESET,
> > @@ -1909,6 +1914,7 @@ static const struct imx_pcie_drvdata drvdata[] = {
> > [IMX8MQ] = {
> > .variant = IMX8MQ,
> > .flags = IMX_PCIE_FLAG_HAS_APP_RESET |
> > + IMX_PCIE_FLAG_KEEP_MSI_CAP |
> > IMX_PCIE_FLAG_HAS_PHY_RESET |
> > IMX_PCIE_FLAG_SUPPORTS_SUSPEND,
> > .gpr = "fsl,imx8mq-iomuxc-gpr",
> > @@ -1923,6 +1929,7 @@ static const struct imx_pcie_drvdata drvdata[] = {
> > [IMX8MM] = {
> > .variant = IMX8MM,
> > .flags = IMX_PCIE_FLAG_SUPPORTS_SUSPEND |
> > + IMX_PCIE_FLAG_KEEP_MSI_CAP |
> > IMX_PCIE_FLAG_HAS_PHYDRV |
> > IMX_PCIE_FLAG_HAS_APP_RESET,
> > .gpr = "fsl,imx8mm-iomuxc-gpr",
> > diff --git a/drivers/pci/controller/dwc/pcie-designware-host.c b/drivers/pci/controller/dwc/pcie-designware-host.c
> > index a74339982c24..7b5558561e15 100644
> > --- a/drivers/pci/controller/dwc/pcie-designware-host.c
> > +++ b/drivers/pci/controller/dwc/pcie-designware-host.c
> > @@ -1171,7 +1171,7 @@ int dw_pcie_setup_rc(struct dw_pcie_rp *pp)
> > * the MSI and MSI-X capabilities of the Root Port to allow the drivers
> > * to fall back to INTx instead.
> > */
> > - if (pp->use_imsi_rx) {
> > + if (pp->use_imsi_rx && !pp->keep_rp_msi_en) {
> > dw_pcie_remove_capability(pci, PCI_CAP_ID_MSI);
> > dw_pcie_remove_capability(pci, PCI_CAP_ID_MSIX);
> > }
> > diff --git a/drivers/pci/controller/dwc/pcie-designware.h b/drivers/pci/controller/dwc/pcie-designware.h
> > index ae6389dd9caa..b12c5334552c 100644
> > --- a/drivers/pci/controller/dwc/pcie-designware.h
> > +++ b/drivers/pci/controller/dwc/pcie-designware.h
> > @@ -421,6 +421,7 @@ struct dw_pcie_host_ops {
> >
> > struct dw_pcie_rp {
> > bool use_imsi_rx:1;
> > + bool keep_rp_msi_en:1;
> > bool cfg0_io_shared:1;
> > u64 cfg0_base;
> > void __iomem *va_cfg0_base;
> > --
> > 2.37.1
> >
--
மணிவண்ணன் சதாசிவம்
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox