* [PATCH 2/2] powerpc/xive: Add a check for memory allocation failure
From: Christophe JAILLET @ 2019-08-01 8:32 UTC (permalink / raw)
To: benh, paulus, mpe, allison, tglx, clg, groug
Cc: kernel-janitors, Christophe JAILLET, linuxppc-dev, linux-kernel
In-Reply-To: <cover.1564647619.git.christophe.jaillet@wanadoo.fr>
The result of this kzalloc is not checked. Add a check and corresponding
error handling code.
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
---
Note that 'xive_irq_bitmap_add()' failures are not handled in
'xive_spapr_init()'
I guess that it is not really an issue. This function is _init, so if a
memory allocation occures here, it is likely that the system will
already be in bad shape.
Anyway, the check added here would at least keep the data linked in
'xive_irq_bitmaps' usable.
---
arch/powerpc/sysdev/xive/spapr.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/arch/powerpc/sysdev/xive/spapr.c b/arch/powerpc/sysdev/xive/spapr.c
index b4f5eb9e0f82..52198131c75e 100644
--- a/arch/powerpc/sysdev/xive/spapr.c
+++ b/arch/powerpc/sysdev/xive/spapr.c
@@ -53,6 +53,10 @@ static int xive_irq_bitmap_add(int base, int count)
xibm->base = base;
xibm->count = count;
xibm->bitmap = kzalloc(xibm->count, GFP_KERNEL);
+ if (!xibm->bitmap) {
+ kfree(xibm);
+ return -ENOMEM;
+ }
list_add(&xibm->list, &xive_irq_bitmaps);
pr_info("Using IRQ range [%x-%x]", xibm->base,
--
2.20.1
^ permalink raw reply related
* Re: [PATCH 1/2] powerpc/xive: Use GFP_KERNEL instead of GFP_ATOMIC in 'xive_irq_bitmap_add()'
From: Cédric Le Goater @ 2019-08-01 8:58 UTC (permalink / raw)
To: Christophe JAILLET, benh, paulus, mpe, allison, tglx, groug
Cc: kernel-janitors, linuxppc-dev, linux-kernel
In-Reply-To: <85d5d247ce753befd6aa63c473f7823de6520ccd.1564647619.git.christophe.jaillet@wanadoo.fr>
On 01/08/2019 10:32, Christophe JAILLET wrote:
> There is no need to use GFP_ATOMIC here. GFP_KERNEL should be enough.
> GFP_KERNEL is also already used for another allocation just a few lines
> below.
This is correct.
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Thanks,
C.
> ---
> arch/powerpc/sysdev/xive/spapr.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/powerpc/sysdev/xive/spapr.c b/arch/powerpc/sysdev/xive/spapr.c
> index 8ef9cf4ebb1c..b4f5eb9e0f82 100644
> --- a/arch/powerpc/sysdev/xive/spapr.c
> +++ b/arch/powerpc/sysdev/xive/spapr.c
> @@ -45,7 +45,7 @@ static int xive_irq_bitmap_add(int base, int count)
> {
> struct xive_irq_bitmap *xibm;
>
> - xibm = kzalloc(sizeof(*xibm), GFP_ATOMIC);
> + xibm = kzalloc(sizeof(*xibm), GFP_KERNEL);
> if (!xibm)
> return -ENOMEM;
>
>
^ permalink raw reply
* Re: [PATCH] powerpc/powernv: Print helpful message when cores guarded
From: Cédric Le Goater @ 2019-08-01 9:33 UTC (permalink / raw)
To: Joel Stanley, Michael Ellerman, linuxppc-dev
In-Reply-To: <20190801051630.5804-1-joel@jms.id.au>
On 01/08/2019 07:16, Joel Stanley wrote:
> Often the firmware will guard out cores after a crash. This often
> undesirable, and is not immediately noticeable.
>
> This adds an informative message when a CPU device tree nodes are marked
> bad in the device tree.
>
> Signed-off-by: Joel Stanley <joel@jms.id.au>
> ---
> Tested on qemu 4.1 with this patch applied:
>
> https://ozlabs.org/~joel/uta2019/0001-TESTING-mark-every-second-core-as-guarded.patch
>
> This will show no cores guarded:
>
> qemu-system-ppc64 -M powernv8 -nographic -smp 1,cores=1,threads=1 -kernel zImage.epapr
>
> This will show three:
>
> qemu-system-ppc64 -M powernv8 -nographic -smp 7,cores=7,threads=1 -kernel zImage.epapr
>
> arch/powerpc/platforms/powernv/setup.c | 24 ++++++++++++++++++++++++
> 1 file changed, 24 insertions(+)
>
> diff --git a/arch/powerpc/platforms/powernv/setup.c b/arch/powerpc/platforms/powernv/setup.c
> index a5e52f9eed3c..7107583d0c6b 100644
> --- a/arch/powerpc/platforms/powernv/setup.c
> +++ b/arch/powerpc/platforms/powernv/setup.c
> @@ -129,6 +129,28 @@ static void pnv_setup_rfi_flush(void)
> setup_count_cache_flush();
> }
>
> +static void __init pnv_check_guarded_cores(void)
> +{
> + struct device_node *dn;
> + int bad_count = 0;
> +
> + for_each_node_by_type(dn, "cpu") {
> + if (of_property_match_string(dn, "status", "bad") >= 0)
> + bad_count++;
> + };
> +
> + if (bad_count) {
> + pr_cont(" __ \n");
> + pr_cont(" / \\ _______________ \n");
> + pr_cont(" | | / \\ \n");
> + pr_cont(" @ @ | WARNING! | \n");
> + pr_cont(" || || | It looks like | \n");
> + pr_cont(" || || <--| you have %*d | \n", 3, bad_count);
> + pr_cont(" |\\_/| | guarded cores | \n");
> + pr_cont(" \\___/ \\_______________/ \n\n");
> + }
> +}
Is that a new animal ? We could use a Cerberus.
Cheers,
C.
> static void __init pnv_setup_arch(void)
> {
> set_arch_panic_timeout(10, ARCH_PANIC_TIMEOUT);
> @@ -149,6 +171,8 @@ static void __init pnv_setup_arch(void)
> /* Enable NAP mode */
> powersave_nap = 1;
>
> + pnv_check_guarded_cores();
> +
> /* XXX PMCS */
> }
>
>
^ permalink raw reply
* Re: [PATCH 1/2] powerpc/xive: Use GFP_KERNEL instead of GFP_ATOMIC in 'xive_irq_bitmap_add()'
From: Greg Kurz @ 2019-08-01 9:23 UTC (permalink / raw)
To: Christophe JAILLET
Cc: kernel-janitors, linux-kernel, paulus, clg, tglx, linuxppc-dev,
allison
In-Reply-To: <85d5d247ce753befd6aa63c473f7823de6520ccd.1564647619.git.christophe.jaillet@wanadoo.fr>
On Thu, 1 Aug 2019 10:32:31 +0200
Christophe JAILLET <christophe.jaillet@wanadoo.fr> wrote:
> There is no need to use GFP_ATOMIC here. GFP_KERNEL should be enough.
> GFP_KERNEL is also already used for another allocation just a few lines
> below.
>
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> ---
Good catch.
Reviewed-by: Greg Kurz <groug@kaod.org>
> arch/powerpc/sysdev/xive/spapr.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/powerpc/sysdev/xive/spapr.c b/arch/powerpc/sysdev/xive/spapr.c
> index 8ef9cf4ebb1c..b4f5eb9e0f82 100644
> --- a/arch/powerpc/sysdev/xive/spapr.c
> +++ b/arch/powerpc/sysdev/xive/spapr.c
> @@ -45,7 +45,7 @@ static int xive_irq_bitmap_add(int base, int count)
> {
> struct xive_irq_bitmap *xibm;
>
> - xibm = kzalloc(sizeof(*xibm), GFP_ATOMIC);
> + xibm = kzalloc(sizeof(*xibm), GFP_KERNEL);
> if (!xibm)
> return -ENOMEM;
>
^ permalink raw reply
* Re: [PATCH 2/2] powerpc/xive: Add a check for memory allocation failure
From: Greg Kurz @ 2019-08-01 9:41 UTC (permalink / raw)
To: Christophe JAILLET
Cc: kernel-janitors, linux-kernel, paulus, clg, tglx, linuxppc-dev,
allison
In-Reply-To: <cc53462734dfeaf15b6bad0e626b483de18656b4.1564647619.git.christophe.jaillet@wanadoo.fr>
On Thu, 1 Aug 2019 10:32:42 +0200
Christophe JAILLET <christophe.jaillet@wanadoo.fr> wrote:
> The result of this kzalloc is not checked. Add a check and corresponding
> error handling code.
>
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> ---
Reviewed-by: Greg Kurz <groug@kaod.org>
> Note that 'xive_irq_bitmap_add()' failures are not handled in
> 'xive_spapr_init()'
> I guess that it is not really an issue. This function is _init, so if a
> memory allocation occures here, it is likely that the system will
> already be in bad shape.
Hmm not sure... The allocation could also fail if the "ibm,xive-lisn-ranges"
property contains an insanely big range, eg. count == 1 << 31. The system isn't
necessarily in bad shape in this case, but XIVE is definitely unusable and
we should let a chance to the kernel to switch to XICS in this case.
I guess it is worth adding proper error handling in xive_spapr_init() as well.
> Anyway, the check added here would at least keep the data linked in
> 'xive_irq_bitmaps' usable.
> ---
> arch/powerpc/sysdev/xive/spapr.c | 4 ++++
> 1 file changed, 4 insertions(+)
>
> diff --git a/arch/powerpc/sysdev/xive/spapr.c b/arch/powerpc/sysdev/xive/spapr.c
> index b4f5eb9e0f82..52198131c75e 100644
> --- a/arch/powerpc/sysdev/xive/spapr.c
> +++ b/arch/powerpc/sysdev/xive/spapr.c
> @@ -53,6 +53,10 @@ static int xive_irq_bitmap_add(int base, int count)
> xibm->base = base;
> xibm->count = count;
> xibm->bitmap = kzalloc(xibm->count, GFP_KERNEL);
> + if (!xibm->bitmap) {
> + kfree(xibm);
> + return -ENOMEM;
> + }
> list_add(&xibm->list, &xive_irq_bitmaps);
>
> pr_info("Using IRQ range [%x-%x]", xibm->base,
^ permalink raw reply
* [PATCH] powerpc/xive: Add some error handling code to 'xive_spapr_init()'
From: Christophe JAILLET @ 2019-08-01 11:09 UTC (permalink / raw)
To: benh, paulus, mpe, allison, tglx, clg, groug
Cc: kernel-janitors, Christophe JAILLET, linuxppc-dev, linux-kernel
'xive_irq_bitmap_add()' can return -ENOMEM.
In this case, we should free the memory already allocated and return
'false' to the caller.
Also add an error path which undoes the 'tima = ioremap(...)'
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
---
NOT compile tested (I don't have a cross compiler and won't install one).
So if some correction or improvement are needed, feel free to propose and
commit it directly.
---
arch/powerpc/sysdev/xive/spapr.c | 36 +++++++++++++++++++++++++-------
1 file changed, 28 insertions(+), 8 deletions(-)
diff --git a/arch/powerpc/sysdev/xive/spapr.c b/arch/powerpc/sysdev/xive/spapr.c
index 52198131c75e..b3ae0b76c433 100644
--- a/arch/powerpc/sysdev/xive/spapr.c
+++ b/arch/powerpc/sysdev/xive/spapr.c
@@ -64,6 +64,17 @@ static int xive_irq_bitmap_add(int base, int count)
return 0;
}
+static void xive_irq_bitmap_remove_all(void)
+{
+ struct xive_irq_bitmap *xibm, *tmp;
+
+ list_for_each_entry_safe(xibm, tmp, &xive_irq_bitmaps, list) {
+ list_del(&xibm->list);
+ kfree(xibm->bitmap);
+ kfree(xibm);
+ }
+}
+
static int __xive_irq_bitmap_alloc(struct xive_irq_bitmap *xibm)
{
int irq;
@@ -723,7 +734,7 @@ bool __init xive_spapr_init(void)
u32 val;
u32 len;
const __be32 *reg;
- int i;
+ int i, err;
if (xive_spapr_disabled())
return false;
@@ -748,23 +759,26 @@ bool __init xive_spapr_init(void)
}
if (!xive_get_max_prio(&max_prio))
- return false;
+ goto err_unmap;
/* Feed the IRQ number allocator with the ranges given in the DT */
reg = of_get_property(np, "ibm,xive-lisn-ranges", &len);
if (!reg) {
pr_err("Failed to read 'ibm,xive-lisn-ranges' property\n");
- return false;
+ goto err_unmap;
}
if (len % (2 * sizeof(u32)) != 0) {
pr_err("invalid 'ibm,xive-lisn-ranges' property\n");
- return false;
+ goto err_unmap;
}
- for (i = 0; i < len / (2 * sizeof(u32)); i++, reg += 2)
- xive_irq_bitmap_add(be32_to_cpu(reg[0]),
- be32_to_cpu(reg[1]));
+ for (i = 0; i < len / (2 * sizeof(u32)); i++, reg += 2) {
+ err = xive_irq_bitmap_add(be32_to_cpu(reg[0]),
+ be32_to_cpu(reg[1]));
+ if (err < 0)
+ goto err_mem_free;
+ }
/* Iterate the EQ sizes and pick one */
of_property_for_each_u32(np, "ibm,xive-eq-sizes", prop, reg, val) {
@@ -775,8 +789,14 @@ bool __init xive_spapr_init(void)
/* Initialize XIVE core with our backend */
if (!xive_core_init(&xive_spapr_ops, tima, TM_QW1_OS, max_prio))
- return false;
+ goto err_mem_free;
pr_info("Using %dkB queues\n", 1 << (xive_queue_shift - 10));
return true;
+
+err_mem_free:
+ xive_irq_bitmap_remove_all();
+err_unmap:
+ iounmap(tima);
+ return false;
}
--
2.20.1
^ permalink raw reply related
* [PATCH] powerpc/mce: Schedule work from irq_work
From: Santosh Sivaraj @ 2019-08-01 11:16 UTC (permalink / raw)
To: linuxppc-dev; +Cc: Mahesh Salgaonkar, Nicholas Piggin
schedule_work() cannot be called from MCE exception context as MCE can
interrupt even in interrupt disabled context.
fixes: 733e4a4c ("powerpc/mce: hookup memory_failure for UE errors")
Signed-off-by: Santosh Sivaraj <santosh@fossix.org>
---
arch/powerpc/kernel/mce.c | 8 +++++---
1 file changed, 5 insertions(+), 3 deletions(-)
diff --git a/arch/powerpc/kernel/mce.c b/arch/powerpc/kernel/mce.c
index b18df633eae9..0ab6fa7cbbbb 100644
--- a/arch/powerpc/kernel/mce.c
+++ b/arch/powerpc/kernel/mce.c
@@ -144,7 +144,6 @@ void save_mce_event(struct pt_regs *regs, long handled,
if (phys_addr != ULONG_MAX) {
mce->u.ue_error.physical_address_provided = true;
mce->u.ue_error.physical_address = phys_addr;
- machine_check_ue_event(mce);
}
}
return;
@@ -275,8 +274,7 @@ static void machine_process_ue_event(struct work_struct *work)
}
}
/*
- * process pending MCE event from the mce event queue. This function will be
- * called during syscall exit.
+ * process pending MCE event from the mce event queue.
*/
static void machine_check_process_queued_event(struct irq_work *work)
{
@@ -292,6 +290,10 @@ static void machine_check_process_queued_event(struct irq_work *work)
while (__this_cpu_read(mce_queue_count) > 0) {
index = __this_cpu_read(mce_queue_count) - 1;
evt = this_cpu_ptr(&mce_event_queue[index]);
+
+ if (evt->error_type == MCE_ERROR_TYPE_UE)
+ machine_check_ue_event(evt);
+
machine_check_print_event_info(evt, false, false);
__this_cpu_dec(mce_queue_count);
}
--
2.20.1
^ permalink raw reply related
* Re: [PATCH] powerpc/xive: Add some error handling code to 'xive_spapr_init()'
From: Cédric Le Goater @ 2019-08-01 11:31 UTC (permalink / raw)
To: Christophe JAILLET, benh, paulus, mpe, allison, tglx, groug
Cc: kernel-janitors, linuxppc-dev, linux-kernel
In-Reply-To: <20190801110956.8517-1-christophe.jaillet@wanadoo.fr>
On 01/08/2019 13:09, Christophe JAILLET wrote:
> 'xive_irq_bitmap_add()' can return -ENOMEM.
> In this case, we should free the memory already allocated and return
> 'false' to the caller.
>
> Also add an error path which undoes the 'tima = ioremap(...)'
>
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> ---
> NOT compile tested (I don't have a cross compiler and won't install one).
All distros have a packaged powerpc cross compiler.
Then, you need to compile a kernel for a pseries machine and run a pseries
machine with it under QEMU. You can use a simple ppc initrd, a net install
one for example.
You could also hack the device tree in QEMU to torture the XIVE sPAPR driver.
Nothing too complex, all is here :
https://git.qemu.org/?p=qemu.git;a=blob;f=hw/intc/spapr_xive.c;h=097f88d4608d8ba160526756a3a224e5176b6e0f;hb=HEAD#l1427
> So if some correction or improvement are needed, feel free to propose and
> commit it directly.
Yes there is I think. I would move at the end all the code that needs a
rollback.
Thanks for taking a look, I might do that one day.
Cheers,
C.
> ---
> arch/powerpc/sysdev/xive/spapr.c | 36 +++++++++++++++++++++++++-------
> 1 file changed, 28 insertions(+), 8 deletions(-)
>
> diff --git a/arch/powerpc/sysdev/xive/spapr.c b/arch/powerpc/sysdev/xive/spapr.c
> index 52198131c75e..b3ae0b76c433 100644
> --- a/arch/powerpc/sysdev/xive/spapr.c
> +++ b/arch/powerpc/sysdev/xive/spapr.c
> @@ -64,6 +64,17 @@ static int xive_irq_bitmap_add(int base, int count)
> return 0;
> }
>
> +static void xive_irq_bitmap_remove_all(void)
> +{
> + struct xive_irq_bitmap *xibm, *tmp;
> +
> + list_for_each_entry_safe(xibm, tmp, &xive_irq_bitmaps, list) {
> + list_del(&xibm->list);
> + kfree(xibm->bitmap);
> + kfree(xibm);
> + }
> +}
> +
> static int __xive_irq_bitmap_alloc(struct xive_irq_bitmap *xibm)
> {
> int irq;
> @@ -723,7 +734,7 @@ bool __init xive_spapr_init(void)
> u32 val;
> u32 len;
> const __be32 *reg;
> - int i;
> + int i, err;
>
> if (xive_spapr_disabled())
> return false;
> @@ -748,23 +759,26 @@ bool __init xive_spapr_init(void)
> }
>
> if (!xive_get_max_prio(&max_prio))
> - return false;
> + goto err_unmap;
>
> /* Feed the IRQ number allocator with the ranges given in the DT */
> reg = of_get_property(np, "ibm,xive-lisn-ranges", &len);
> if (!reg) {
> pr_err("Failed to read 'ibm,xive-lisn-ranges' property\n");
> - return false;
> + goto err_unmap;
> }
>
> if (len % (2 * sizeof(u32)) != 0) {
> pr_err("invalid 'ibm,xive-lisn-ranges' property\n");
> - return false;
> + goto err_unmap;
> }
>
> - for (i = 0; i < len / (2 * sizeof(u32)); i++, reg += 2)
> - xive_irq_bitmap_add(be32_to_cpu(reg[0]),
> - be32_to_cpu(reg[1]));
> + for (i = 0; i < len / (2 * sizeof(u32)); i++, reg += 2) {
> + err = xive_irq_bitmap_add(be32_to_cpu(reg[0]),
> + be32_to_cpu(reg[1]));
> + if (err < 0)
> + goto err_mem_free;
> + }
>
> /* Iterate the EQ sizes and pick one */
> of_property_for_each_u32(np, "ibm,xive-eq-sizes", prop, reg, val) {
> @@ -775,8 +789,14 @@ bool __init xive_spapr_init(void)
>
> /* Initialize XIVE core with our backend */
> if (!xive_core_init(&xive_spapr_ops, tima, TM_QW1_OS, max_prio))
> - return false;
> + goto err_mem_free;
>
> pr_info("Using %dkB queues\n", 1 << (xive_queue_shift - 10));
> return true;
> +
> +err_mem_free:
> + xive_irq_bitmap_remove_all();
> +err_unmap:
> + iounmap(tima);
> + return false;
> }
>
^ permalink raw reply
* [Bug 204375] kernel 5.2.4 w. KASAN enabled fails to boot on a PowerMac G4 3,6 at very early stage
From: bugzilla-daemon @ 2019-08-01 11:53 UTC (permalink / raw)
To: linuxppc-dev
In-Reply-To: <bug-204375-206035@https.bugzilla.kernel.org/>
https://bugzilla.kernel.org/show_bug.cgi?id=204375
--- Comment #13 from Erhard F. (erhard_f@mailbox.org) ---
Thanks for the hint! But adding KASAN_SANITIZE := n to lib/mpi/Makefile did not
make a change in this case.
However I had the idea to disable btrfs (at all) and boot from my other ext4
partition. This worked out and I got further. After other stacktraces flying by
the boot process now stalls at offb switching over to radeondrmfb. Again reboot
after timer kicks in after 10min.
Seems btrfs is not in the best condition on ppc32 with 4K pagesize..
I would test this on the G5 or Talos II too, but there is no kernel option to
enable 'KASAN: runtime memory debugging' on ppc64.
--
You are receiving this mail because:
You are watching the assignee of the bug.
^ permalink raw reply
* [PATCH] selftests/powerpc: Fix build failures with GCC 9
From: Michael Ellerman @ 2019-08-01 12:26 UTC (permalink / raw)
To: linuxppc-dev
GCC 9 fails to build some of the ptrace TM tests, with errors such as:
ptrace-tm-spd-vsx.c: In function 'tm_spd_vsx':
ptrace-tm-spd-vsx.c:51:2: error: listing the stack pointer register 'r1' in a clobber list is deprecated [-Werror=deprecated]
51 | asm __volatile__(
| ^~~
ptrace-tm-spd-vsx.c:51:2: note: the value of the stack pointer after an 'asm' statement must be the same as it was before the statement
Which is probably fair enough.
Some of these inline asm blocks are doing quite a lot and are probably
pushing the boundaries of what's sane to do with inline asm, but they
shouldn't actually be returning with r1 modified.
So drop r1 from the clobbers for now, we should probably rewrite them
to be real asm functions at some point.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
---
tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-tar.c | 2 +-
tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-vsx.c | 3 +--
tools/testing/selftests/powerpc/ptrace/ptrace-tm-tar.c | 2 +-
tools/testing/selftests/powerpc/ptrace/ptrace-tm-vsx.c | 3 +--
4 files changed, 4 insertions(+), 6 deletions(-)
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-tar.c b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-tar.c
index 25e23e73c72e..7b835ef4f8a6 100644
--- a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-tar.c
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-tar.c
@@ -73,7 +73,7 @@ void tm_spd_tar(void)
[sprn_texasr]"i"(SPRN_TEXASR), [tar_1]"i"(TAR_1),
[dscr_1]"i"(DSCR_1), [tar_2]"i"(TAR_2), [dscr_2]"i"(DSCR_2),
[tar_3]"i"(TAR_3), [dscr_3]"i"(DSCR_3)
- : "memory", "r0", "r1", "r3", "r4", "r5", "r6"
+ : "memory", "r0", "r3", "r4", "r5", "r6"
);
/* TM failed, analyse */
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-vsx.c b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-vsx.c
index f603fe5a445b..724e5aa499cd 100644
--- a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-vsx.c
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-vsx.c
@@ -74,8 +74,7 @@ void tm_spd_vsx(void)
"3: ;"
: [res] "=r" (result), [texasr] "=r" (texasr)
: [sprn_texasr] "i" (SPRN_TEXASR)
- : "memory", "r0", "r1", "r3", "r4",
- "r7", "r8", "r9", "r10", "r11"
+ : "memory", "r0", "r3", "r4", "r7", "r8", "r9", "r10", "r11"
);
if (result) {
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-tar.c b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-tar.c
index e0d37f07bdeb..46ef378a15ec 100644
--- a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-tar.c
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-tar.c
@@ -62,7 +62,7 @@ void tm_tar(void)
[sprn_ppr]"i"(SPRN_PPR), [sprn_texasr]"i"(SPRN_TEXASR),
[tar_1]"i"(TAR_1), [dscr_1]"i"(DSCR_1), [tar_2]"i"(TAR_2),
[dscr_2]"i"(DSCR_2), [cptr1] "b" (&cptr[1])
- : "memory", "r0", "r1", "r3", "r4", "r5", "r6"
+ : "memory", "r0", "r3", "r4", "r5", "r6"
);
/* TM failed, analyse */
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-vsx.c b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-vsx.c
index 8027457b97b7..9f16f3a74e28 100644
--- a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-vsx.c
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-vsx.c
@@ -62,8 +62,7 @@ void tm_vsx(void)
"3: ;"
: [res] "=r" (result), [texasr] "=r" (texasr)
: [sprn_texasr] "i" (SPRN_TEXASR), [cptr1] "b" (&cptr[1])
- : "memory", "r0", "r1", "r3", "r4",
- "r7", "r8", "r9", "r10", "r11"
+ : "memory", "r0", "r3", "r4", "r7", "r8", "r9", "r10", "r11"
);
if (result) {
--
2.21.0
^ permalink raw reply related
* [Bug 204371] BUG kmalloc-4k (Tainted: G W ): Object padding overwritten
From: bugzilla-daemon @ 2019-08-01 12:47 UTC (permalink / raw)
To: linuxppc-dev
In-Reply-To: <bug-204371-206035@https.bugzilla.kernel.org/>
https://bugzilla.kernel.org/show_bug.cgi?id=204371
--- Comment #5 from Erhard F. (erhard_f@mailbox.org) ---
On Wed, 31 Jul 2019 12:09:54 +0000
bugzilla-daemon@bugzilla.kernel.org wrote:
> https://bugzilla.kernel.org/show_bug.cgi?id=204371
>
> --- Comment #4 from mpe@ellerman.id.au ---
>
> > I suspect proc_cgroup_show() is innocent and that perhaps
> > bpf_prepare_filter() had a memory scribble. iirc there has been at
> > least one recent pretty serious bpf fix applied recently. Can others
> > please take a look?
>
> I haven't been able to reproduce this on a 64-bit or 32-bit powerpc
> machine here. But I don't run gentoo userspace, so I suspect I'm not
> tripping the same path at boot. I did run the seccomp selftest and that
> didn't trip it either.
>
> cheers
Doing some fiddling around on another bug (bug #204375), I noticed that I get
this "kmalloc-4k (Tainted: G W ): Object padding overwritten" during boot only
when I boot from my btrfs partition, but not from my other ext4 partition. The
ext4 partition is not a clone, but pretty much the same stuff in the same
versions. My btrfs root is mounted with 'lazytime,compress=zstd:1', systemd is
242.
I built a 5.2.5 kernel on the Talos II with CONFIG_SLUB_DEBUG=y but here I
don't hit the bug, even if I boot from a btrfs partition with the same
settings. Have to test it on the G5 yet (kernel .config more similar to the G4
one than the Talos II one).
--
You are receiving this mail because:
You are on the CC list for the bug.
^ permalink raw reply
* Re: [PATCH 6/8] dma-direct: turn ARCH_ZONE_DMA_BITS into a variable
From: Christoph Hellwig @ 2019-08-01 14:04 UTC (permalink / raw)
To: Nicolas Saenz Julienne
Cc: Heiko Carstens, eric, Paul Mackerras, will, hch, Marek Szyprowski,
phill, linux-s390, f.fainelli, frowand.list,
Christian Borntraeger, catalin.marinas, devicetree, Vasily Gorbik,
marc.zyngier, robh+dt, linux-rpi-kernel, linux-arm-kernel,
mbrugger, linux-mm, linuxppc-dev, linux-kernel, iommu, wahrenst,
akpm, Robin Murphy
In-Reply-To: <20190731154752.16557-7-nsaenzjulienne@suse.de>
A few nitpicks, otherwise this looks great:
> @@ -201,7 +202,7 @@ static int __init mark_nonram_nosave(void)
> * everything else. GFP_DMA32 page allocations automatically fall back to
> * ZONE_DMA.
> *
> - * By using 31-bit unconditionally, we can exploit ARCH_ZONE_DMA_BITS to
> + * By using 31-bit unconditionally, we can exploit arch_zone_dma_bits to
> * inform the generic DMA mapping code. 32-bit only devices (if not handled
> * by an IOMMU anyway) will take a first dip into ZONE_NORMAL and get
> * otherwise served by ZONE_DMA.
> @@ -237,9 +238,18 @@ void __init paging_init(void)
> printk(KERN_DEBUG "Memory hole size: %ldMB\n",
> (long int)((top_of_ram - total_ram) >> 20));
>
> + /*
> + * Allow 30-bit DMA for very limited Broadcom wifi chips on many
> + * powerbooks.
> + */
> + if (IS_ENABLED(CONFIG_PPC32))
> + arch_zone_dma_bits = 30;
> + else
> + arch_zone_dma_bits = 31;
> +
So the above unconditionally comment obviously isn't true any more, and
Ben also said for the recent ppc32 hack he'd prefer dynamic detection.
Maybe Ben and or other ppc folks can chime in an add a patch to the series
to sort this out now that we have a dynamic ZONE_DMA threshold?
> diff --git a/kernel/dma/direct.c b/kernel/dma/direct.c
> index 59bdceea3737..40dfc9b4ee4c 100644
> --- a/kernel/dma/direct.c
> +++ b/kernel/dma/direct.c
> @@ -19,9 +19,7 @@
> * Most architectures use ZONE_DMA for the first 16 Megabytes, but
> * some use it for entirely different regions:
> */
> -#ifndef ARCH_ZONE_DMA_BITS
> -#define ARCH_ZONE_DMA_BITS 24
> -#endif
> +unsigned int arch_zone_dma_bits __ro_after_init = 24;
I'd prefer to drop the arch_ prefix and just calls this zone_dma_bits.
In the long run we really need to find a way to just automatically set
this from the meminit code, but that is out of scope for this series.
For now can you please just update the comment above to say something
like:
/*
* Most architectures use ZONE_DMA for the first 16 Megabytes, but some use it
* it for entirely different regions. In that case the arch code needs to
* override the variable below for dma-direct to work properly.
*/
^ permalink raw reply
* fix default dma_mmap_* pgprot
From: Christoph Hellwig @ 2019-08-01 14:21 UTC (permalink / raw)
To: iommu
Cc: Shawn Anastasio, Will Deacon, linuxppc-dev, Russell King,
linux-kernel, Catalin Marinas, Robin Murphy, linux-arm-kernel
Hi all,
As Shawn pointed out we've had issues with the dma mmap pgprots ever
since the dma_common_mmap helper was added beyong the initial
architectures - we default to uncached mappings, but for devices that
are DMA coherent, or if the DMA_ATTR_NON_CONSISTENT is set (and
supported) this can lead to aliasing of cache attributes. This patch
fixes that. My explanation of why this hasn't been much of an issue
is that the dma_mmap_ helpers aren't used widely and mostly just in
architecture specific drivers.
^ permalink raw reply
* [PATCH] dma-mapping: fix page attributes for dma_mmap_*
From: Christoph Hellwig @ 2019-08-01 14:21 UTC (permalink / raw)
To: iommu
Cc: Shawn Anastasio, Will Deacon, linuxppc-dev, Russell King,
linux-kernel, Catalin Marinas, Robin Murphy, linux-arm-kernel
In-Reply-To: <20190801142118.21225-1-hch@lst.de>
All the way back to introducing dma_common_mmap we've defaulyed to mark
the pages as uncached. But this is wrong for DMA coherent devices or
if using DMA_ATTR_NON_CONSISTENT. Later on DMA_ATTR_WRITE_COMBINE
also got incorrect treatment as that flag is only treated special on
the alloc side for non-coherent devices.
Introduce a new dma_mmap_pgprot helper that deals with the check
for coherent devices and DMA_ATTR_NON_CONSISTENT so that only the
remapping cases even reach arch_dma_mmap_pgprot and we thus ensure
no aliasing of page attributes happens.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
arch/arm/mm/dma-mapping.c | 4 +---
arch/arm64/mm/dma-mapping.c | 4 +---
arch/powerpc/kernel/Makefile | 3 +--
arch/powerpc/kernel/dma-common.c | 17 -----------------
drivers/iommu/dma-iommu.c | 6 +++---
include/linux/dma-mapping.h | 1 +
include/linux/dma-noncoherent.h | 5 -----
kernel/dma/mapping.c | 11 ++++++++++-
kernel/dma/remap.c | 2 +-
9 files changed, 18 insertions(+), 35 deletions(-)
delete mode 100644 arch/powerpc/kernel/dma-common.c
diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c
index 9c9a23e5600d..cfe44df169c5 100644
--- a/arch/arm/mm/dma-mapping.c
+++ b/arch/arm/mm/dma-mapping.c
@@ -2397,9 +2397,7 @@ long arch_dma_coherent_to_pfn(struct device *dev, void *cpu_addr,
pgprot_t arch_dma_mmap_pgprot(struct device *dev, pgprot_t prot,
unsigned long attrs)
{
- if (!dev_is_dma_coherent(dev))
- return __get_dma_pgprot(attrs, prot);
- return prot;
+ return __get_dma_pgprot(attrs, prot);
}
void *arch_dma_alloc(struct device *dev, size_t size, dma_addr_t *dma_handle,
diff --git a/arch/arm64/mm/dma-mapping.c b/arch/arm64/mm/dma-mapping.c
index 1d3f0b5a9940..bd2b039f43a6 100644
--- a/arch/arm64/mm/dma-mapping.c
+++ b/arch/arm64/mm/dma-mapping.c
@@ -14,9 +14,7 @@
pgprot_t arch_dma_mmap_pgprot(struct device *dev, pgprot_t prot,
unsigned long attrs)
{
- if (!dev_is_dma_coherent(dev) || (attrs & DMA_ATTR_WRITE_COMBINE))
- return pgprot_writecombine(prot);
- return prot;
+ return pgprot_writecombine(prot);
}
void arch_sync_dma_for_device(struct device *dev, phys_addr_t paddr,
diff --git a/arch/powerpc/kernel/Makefile b/arch/powerpc/kernel/Makefile
index ea0c69236789..56dfa7a2a6f2 100644
--- a/arch/powerpc/kernel/Makefile
+++ b/arch/powerpc/kernel/Makefile
@@ -49,8 +49,7 @@ obj-y := cputable.o ptrace.o syscalls.o \
signal.o sysfs.o cacheinfo.o time.o \
prom.o traps.o setup-common.o \
udbg.o misc.o io.o misc_$(BITS).o \
- of_platform.o prom_parse.o \
- dma-common.o
+ of_platform.o prom_parse.o
obj-$(CONFIG_PPC64) += setup_64.o sys_ppc32.o \
signal_64.o ptrace32.o \
paca.o nvram_64.o firmware.o
diff --git a/arch/powerpc/kernel/dma-common.c b/arch/powerpc/kernel/dma-common.c
deleted file mode 100644
index dc7ef6b17b69..000000000000
--- a/arch/powerpc/kernel/dma-common.c
+++ /dev/null
@@ -1,17 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0-or-later
-/*
- * Contains common dma routines for all powerpc platforms.
- *
- * Copyright (C) 2019 Shawn Anastasio.
- */
-
-#include <linux/mm.h>
-#include <linux/dma-noncoherent.h>
-
-pgprot_t arch_dma_mmap_pgprot(struct device *dev, pgprot_t prot,
- unsigned long attrs)
-{
- if (!dev_is_dma_coherent(dev))
- return pgprot_noncached(prot);
- return prot;
-}
diff --git a/drivers/iommu/dma-iommu.c b/drivers/iommu/dma-iommu.c
index a7f9c3edbcb2..703de9a7553f 100644
--- a/drivers/iommu/dma-iommu.c
+++ b/drivers/iommu/dma-iommu.c
@@ -574,7 +574,7 @@ static void *iommu_dma_alloc_remap(struct device *dev, size_t size,
struct iova_domain *iovad = &cookie->iovad;
bool coherent = dev_is_dma_coherent(dev);
int ioprot = dma_info_to_prot(DMA_BIDIRECTIONAL, coherent, attrs);
- pgprot_t prot = arch_dma_mmap_pgprot(dev, PAGE_KERNEL, attrs);
+ pgprot_t prot = dma_mmap_pgprot(dev, PAGE_KERNEL, attrs);
unsigned int count, min_size, alloc_sizes = domain->pgsize_bitmap;
struct page **pages;
struct sg_table sgt;
@@ -975,7 +975,7 @@ static void *iommu_dma_alloc_pages(struct device *dev, size_t size,
return NULL;
if (IS_ENABLED(CONFIG_DMA_REMAP) && (!coherent || PageHighMem(page))) {
- pgprot_t prot = arch_dma_mmap_pgprot(dev, PAGE_KERNEL, attrs);
+ pgprot_t prot = dma_mmap_pgprot(dev, PAGE_KERNEL, attrs);
cpu_addr = dma_common_contiguous_remap(page, alloc_size,
VM_USERMAP, prot, __builtin_return_address(0));
@@ -1035,7 +1035,7 @@ static int iommu_dma_mmap(struct device *dev, struct vm_area_struct *vma,
unsigned long pfn, off = vma->vm_pgoff;
int ret;
- vma->vm_page_prot = arch_dma_mmap_pgprot(dev, vma->vm_page_prot, attrs);
+ vma->vm_page_prot = dma_mmap_pgprot(dev, vma->vm_page_prot, attrs);
if (dma_mmap_from_dev_coherent(dev, vma, cpu_addr, size, &ret))
return ret;
diff --git a/include/linux/dma-mapping.h b/include/linux/dma-mapping.h
index 633dae466097..c61d7870277f 100644
--- a/include/linux/dma-mapping.h
+++ b/include/linux/dma-mapping.h
@@ -611,6 +611,7 @@ static inline void dma_sync_single_range_for_device(struct device *dev,
#define dma_get_sgtable(d, t, v, h, s) dma_get_sgtable_attrs(d, t, v, h, s, 0)
#define dma_mmap_coherent(d, v, c, h, s) dma_mmap_attrs(d, v, c, h, s, 0)
+pgprot_t dma_mmap_pgprot(struct device *dev, pgprot_t prot, unsigned long attrs);
extern int dma_common_mmap(struct device *dev, struct vm_area_struct *vma,
void *cpu_addr, dma_addr_t dma_addr, size_t size,
unsigned long attrs);
diff --git a/include/linux/dma-noncoherent.h b/include/linux/dma-noncoherent.h
index 3813211a9aad..9ae5cee543c4 100644
--- a/include/linux/dma-noncoherent.h
+++ b/include/linux/dma-noncoherent.h
@@ -42,13 +42,8 @@ void arch_dma_free(struct device *dev, size_t size, void *cpu_addr,
dma_addr_t dma_addr, unsigned long attrs);
long arch_dma_coherent_to_pfn(struct device *dev, void *cpu_addr,
dma_addr_t dma_addr);
-
-#ifdef CONFIG_ARCH_HAS_DMA_MMAP_PGPROT
pgprot_t arch_dma_mmap_pgprot(struct device *dev, pgprot_t prot,
unsigned long attrs);
-#else
-# define arch_dma_mmap_pgprot(dev, prot, attrs) pgprot_noncached(prot)
-#endif
#ifdef CONFIG_DMA_NONCOHERENT_CACHE_SYNC
void arch_dma_cache_sync(struct device *dev, void *vaddr, size_t size,
diff --git a/kernel/dma/mapping.c b/kernel/dma/mapping.c
index 815446f76995..357ae5cdb91b 100644
--- a/kernel/dma/mapping.c
+++ b/kernel/dma/mapping.c
@@ -162,6 +162,15 @@ int dma_get_sgtable_attrs(struct device *dev, struct sg_table *sgt,
}
EXPORT_SYMBOL(dma_get_sgtable_attrs);
+pgprot_t dma_mmap_pgprot(struct device *dev, pgprot_t prot, unsigned long attrs)
+{
+ if (dev_is_dma_coherent(dev) || (attrs & DMA_ATTR_NON_CONSISTENT))
+ return prot;
+ if (IS_ENABLED(CONFIG_ARCH_HAS_DMA_MMAP_PGPROT))
+ return arch_dma_mmap_pgprot(dev, prot, attrs);
+ return pgprot_noncached(prot);
+}
+
/*
* Create userspace mapping for the DMA-coherent memory.
*/
@@ -175,7 +184,7 @@ int dma_common_mmap(struct device *dev, struct vm_area_struct *vma,
unsigned long pfn;
int ret = -ENXIO;
- vma->vm_page_prot = arch_dma_mmap_pgprot(dev, vma->vm_page_prot, attrs);
+ vma->vm_page_prot = dma_mmap_pgprot(dev, vma->vm_page_prot, attrs);
if (dma_mmap_from_dev_coherent(dev, vma, cpu_addr, size, &ret))
return ret;
diff --git a/kernel/dma/remap.c b/kernel/dma/remap.c
index a594aec07882..3d75c79b124c 100644
--- a/kernel/dma/remap.c
+++ b/kernel/dma/remap.c
@@ -218,7 +218,7 @@ void *arch_dma_alloc(struct device *dev, size_t size, dma_addr_t *dma_handle,
/* create a coherent mapping */
ret = dma_common_contiguous_remap(page, size, VM_USERMAP,
- arch_dma_mmap_pgprot(dev, PAGE_KERNEL, attrs),
+ dma_mmap_pgprot(dev, PAGE_KERNEL, attrs),
__builtin_return_address(0));
if (!ret) {
__dma_direct_free_pages(dev, size, page);
--
2.20.1
^ permalink raw reply related
* Re: [PATCH v3 00/10] implement KASLR for powerpc/fsl_booke/32
From: Diana Madalina Craciun @ 2019-08-01 14:36 UTC (permalink / raw)
To: Jason Yan, mpe@ellerman.id.au, linuxppc-dev@lists.ozlabs.org,
christophe.leroy@c-s.fr, benh@kernel.crashing.org,
paulus@samba.org, npiggin@gmail.com, keescook@chromium.org,
kernel-hardening@lists.openwall.com
Cc: wangkefeng.wang@huawei.com, linux-kernel@vger.kernel.org,
jingxiangfeng@huawei.com, zhaohongjiang@huawei.com,
thunder.leizhen@huawei.com, fanchengyang@huawei.com,
yebin10@huawei.com
In-Reply-To: <20190731094318.26538-1-yanaijie@huawei.com>
Hi Jason,
I have tested these series on a P4080 platform.
Regards,
Diana
On 7/31/2019 12:26 PM, Jason Yan wrote:
> This series implements KASLR for powerpc/fsl_booke/32, as a security
> feature that deters exploit attempts relying on knowledge of the location
> of kernel internals.
>
> Since CONFIG_RELOCATABLE has already supported, what we need to do is
> map or copy kernel to a proper place and relocate. Freescale Book-E
> parts expect lowmem to be mapped by fixed TLB entries(TLB1). The TLB1
> entries are not suitable to map the kernel directly in a randomized
> region, so we chose to copy the kernel to a proper place and restart to
> relocate.
>
> Entropy is derived from the banner and timer base, which will change every
> build and boot. This not so much safe so additionally the bootloader may
> pass entropy via the /chosen/kaslr-seed node in device tree.
>
> We will use the first 512M of the low memory to randomize the kernel
> image. The memory will be split in 64M zones. We will use the lower 8
> bit of the entropy to decide the index of the 64M zone. Then we chose a
> 16K aligned offset inside the 64M zone to put the kernel in.
>
> KERNELBASE
>
> |--> 64M <--|
> | |
> +---------------+ +----------------+---------------+
> | |....| |kernel| | |
> +---------------+ +----------------+---------------+
> | |
> |-----> offset <-----|
>
> kimage_vaddr
>
> We also check if we will overlap with some areas like the dtb area, the
> initrd area or the crashkernel area. If we cannot find a proper area,
> kaslr will be disabled and boot from the original kernel.
>
> Changes since v2:
> - Remove unnecessary #ifdef
> - Use SZ_64M instead of0x4000000
> - Call early_init_dt_scan_chosen() to init boot_command_line
> - Rename kaslr_second_init() to kaslr_late_init()
>
> Changes since v1:
> - Remove some useless 'extern' keyword.
> - Replace EXPORT_SYMBOL with EXPORT_SYMBOL_GPL
> - Improve some assembly code
> - Use memzero_explicit instead of memset
> - Use boot_command_line and remove early_command_line
> - Do not print kaslr offset if kaslr is disabled
>
> Jason Yan (10):
> powerpc: unify definition of M_IF_NEEDED
> powerpc: move memstart_addr and kernstart_addr to init-common.c
> powerpc: introduce kimage_vaddr to store the kernel base
> powerpc/fsl_booke/32: introduce create_tlb_entry() helper
> powerpc/fsl_booke/32: introduce reloc_kernel_entry() helper
> powerpc/fsl_booke/32: implement KASLR infrastructure
> powerpc/fsl_booke/32: randomize the kernel image offset
> powerpc/fsl_booke/kaslr: clear the original kernel if randomized
> powerpc/fsl_booke/kaslr: support nokaslr cmdline parameter
> powerpc/fsl_booke/kaslr: dump out kernel offset information on panic
>
> arch/powerpc/Kconfig | 11 +
> arch/powerpc/include/asm/nohash/mmu-book3e.h | 10 +
> arch/powerpc/include/asm/page.h | 7 +
> arch/powerpc/kernel/Makefile | 1 +
> arch/powerpc/kernel/early_32.c | 2 +-
> arch/powerpc/kernel/exceptions-64e.S | 10 -
> arch/powerpc/kernel/fsl_booke_entry_mapping.S | 23 +-
> arch/powerpc/kernel/head_fsl_booke.S | 55 ++-
> arch/powerpc/kernel/kaslr_booke.c | 427 ++++++++++++++++++
> arch/powerpc/kernel/machine_kexec.c | 1 +
> arch/powerpc/kernel/misc_64.S | 5 -
> arch/powerpc/kernel/setup-common.c | 19 +
> arch/powerpc/mm/init-common.c | 7 +
> arch/powerpc/mm/init_32.c | 5 -
> arch/powerpc/mm/init_64.c | 5 -
> arch/powerpc/mm/mmu_decl.h | 10 +
> arch/powerpc/mm/nohash/fsl_booke.c | 8 +-
> 17 files changed, 558 insertions(+), 48 deletions(-)
> create mode 100644 arch/powerpc/kernel/kaslr_booke.c
>
^ permalink raw reply
* Re: [PATCH 6/8] dma-direct: turn ARCH_ZONE_DMA_BITS into a variable
From: Nicolas Saenz Julienne @ 2019-08-01 15:59 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Heiko Carstens, eric, Paul Mackerras, will, Marek Szyprowski,
phill, linux-s390, f.fainelli, frowand.list,
Christian Borntraeger, catalin.marinas, devicetree, Vasily Gorbik,
marc.zyngier, robh+dt, linux-rpi-kernel, linux-arm-kernel,
mbrugger, linux-mm, linuxppc-dev, linux-kernel, iommu, wahrenst,
akpm, Robin Murphy
In-Reply-To: <20190801140452.GB23435@lst.de>
[-- Attachment #1: Type: text/plain, Size: 2485 bytes --]
Hi Christoph, thanks for the review.
On Thu, 2019-08-01 at 16:04 +0200, Christoph Hellwig wrote:
> A few nitpicks, otherwise this looks great:
>
> > @@ -201,7 +202,7 @@ static int __init mark_nonram_nosave(void)
> > * everything else. GFP_DMA32 page allocations automatically fall back to
> > * ZONE_DMA.
> > *
> > - * By using 31-bit unconditionally, we can exploit ARCH_ZONE_DMA_BITS to
> > + * By using 31-bit unconditionally, we can exploit arch_zone_dma_bits to
> > * inform the generic DMA mapping code. 32-bit only devices (if not
> > handled
> > * by an IOMMU anyway) will take a first dip into ZONE_NORMAL and get
> > * otherwise served by ZONE_DMA.
> > @@ -237,9 +238,18 @@ void __init paging_init(void)
> > printk(KERN_DEBUG "Memory hole size: %ldMB\n",
> > (long int)((top_of_ram - total_ram) >> 20));
> >
> > + /*
> > + * Allow 30-bit DMA for very limited Broadcom wifi chips on many
> > + * powerbooks.
> > + */
> > + if (IS_ENABLED(CONFIG_PPC32))
> > + arch_zone_dma_bits = 30;
> > + else
> > + arch_zone_dma_bits = 31;
> > +
>
> So the above unconditionally comment obviously isn't true any more, and
> Ben also said for the recent ppc32 hack he'd prefer dynamic detection.
>
> Maybe Ben and or other ppc folks can chime in an add a patch to the series
> to sort this out now that we have a dynamic ZONE_DMA threshold?
Noted, for now I'll remove the comment.
> > diff --git a/kernel/dma/direct.c b/kernel/dma/direct.c
> > index 59bdceea3737..40dfc9b4ee4c 100644
> > --- a/kernel/dma/direct.c
> > +++ b/kernel/dma/direct.c
> > @@ -19,9 +19,7 @@
> > * Most architectures use ZONE_DMA for the first 16 Megabytes, but
> > * some use it for entirely different regions:
> > */
> > -#ifndef ARCH_ZONE_DMA_BITS
> > -#define ARCH_ZONE_DMA_BITS 24
> > -#endif
> > +unsigned int arch_zone_dma_bits __ro_after_init = 24;
>
> I'd prefer to drop the arch_ prefix and just calls this zone_dma_bits.
> In the long run we really need to find a way to just automatically set
> this from the meminit code, but that is out of scope for this series.
> For now can you please just update the comment above to say something
> like:
>
> /*
> * Most architectures use ZONE_DMA for the first 16 Megabytes, but some use it
> * it for entirely different regions. In that case the arch code needs to
> * override the variable below for dma-direct to work properly.
> */
Ok perfect.
[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 488 bytes --]
^ permalink raw reply
* Re: [PATCH] dma-mapping: fix page attributes for dma_mmap_*
From: Will Deacon @ 2019-08-01 16:23 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Shawn Anastasio, linuxppc-dev, Russell King, linux-kernel, iommu,
Catalin Marinas, Robin Murphy, linux-arm-kernel
In-Reply-To: <20190801142118.21225-2-hch@lst.de>
On Thu, Aug 01, 2019 at 05:21:18PM +0300, Christoph Hellwig wrote:
> All the way back to introducing dma_common_mmap we've defaulyed to mark
> the pages as uncached. But this is wrong for DMA coherent devices or
> if using DMA_ATTR_NON_CONSISTENT. Later on DMA_ATTR_WRITE_COMBINE
> also got incorrect treatment as that flag is only treated special on
> the alloc side for non-coherent devices.
>
> Introduce a new dma_mmap_pgprot helper that deals with the check
> for coherent devices and DMA_ATTR_NON_CONSISTENT so that only the
> remapping cases even reach arch_dma_mmap_pgprot and we thus ensure
> no aliasing of page attributes happens.
>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> ---
> arch/arm/mm/dma-mapping.c | 4 +---
> arch/arm64/mm/dma-mapping.c | 4 +---
> arch/powerpc/kernel/Makefile | 3 +--
> arch/powerpc/kernel/dma-common.c | 17 -----------------
> drivers/iommu/dma-iommu.c | 6 +++---
> include/linux/dma-mapping.h | 1 +
> include/linux/dma-noncoherent.h | 5 -----
> kernel/dma/mapping.c | 11 ++++++++++-
> kernel/dma/remap.c | 2 +-
> 9 files changed, 18 insertions(+), 35 deletions(-)
> delete mode 100644 arch/powerpc/kernel/dma-common.c
>
> diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c
> index 9c9a23e5600d..cfe44df169c5 100644
> --- a/arch/arm/mm/dma-mapping.c
> +++ b/arch/arm/mm/dma-mapping.c
> @@ -2397,9 +2397,7 @@ long arch_dma_coherent_to_pfn(struct device *dev, void *cpu_addr,
> pgprot_t arch_dma_mmap_pgprot(struct device *dev, pgprot_t prot,
> unsigned long attrs)
> {
> - if (!dev_is_dma_coherent(dev))
> - return __get_dma_pgprot(attrs, prot);
> - return prot;
> + return __get_dma_pgprot(attrs, prot);
> }
>
> void *arch_dma_alloc(struct device *dev, size_t size, dma_addr_t *dma_handle,
> diff --git a/arch/arm64/mm/dma-mapping.c b/arch/arm64/mm/dma-mapping.c
> index 1d3f0b5a9940..bd2b039f43a6 100644
> --- a/arch/arm64/mm/dma-mapping.c
> +++ b/arch/arm64/mm/dma-mapping.c
> @@ -14,9 +14,7 @@
> pgprot_t arch_dma_mmap_pgprot(struct device *dev, pgprot_t prot,
> unsigned long attrs)
> {
> - if (!dev_is_dma_coherent(dev) || (attrs & DMA_ATTR_WRITE_COMBINE))
> - return pgprot_writecombine(prot);
> - return prot;
> + return pgprot_writecombine(prot);
> }
Seems like a sensible cleanup to me:
Acked-by: Will Deacon <will@kernel.org>
Although arch_dma_mmap_pgprot() is a bit of a misnomer now that it only
gets involved in the non-coherent case.
Will
^ permalink raw reply
* Re: [PATCH] dma-mapping: fix page attributes for dma_mmap_*
From: Christoph Hellwig @ 2019-08-01 16:34 UTC (permalink / raw)
To: Will Deacon
Cc: Shawn Anastasio, linuxppc-dev, Russell King, linux-kernel, iommu,
Catalin Marinas, Robin Murphy, Christoph Hellwig,
linux-arm-kernel
In-Reply-To: <20190801162305.3m32chycsdjmdejk@willie-the-truck>
On Thu, Aug 01, 2019 at 05:23:06PM +0100, Will Deacon wrote:
> > - if (!dev_is_dma_coherent(dev) || (attrs & DMA_ATTR_WRITE_COMBINE))
> > - return pgprot_writecombine(prot);
> > - return prot;
> > + return pgprot_writecombine(prot);
> > }
>
> Seems like a sensible cleanup to me:
>
> Acked-by: Will Deacon <will@kernel.org>
>
> Although arch_dma_mmap_pgprot() is a bit of a misnomer now that it only
> gets involved in the non-coherent case.
A better name is welcome. My other idea would be to just remove it
entirely and do something like:
#ifndef pgprot_dmacoherent
#define pgprot_dmacoherent pgprot_noncached
#endif
pgprot_t dma_mmap_pgprot(struct device *dev, pgprot_t prot, unsigned long attrs)
{
if (dev_is_dma_coherent(dev) || (attrs & DMA_ATTR_NON_CONSISTENT))
return prot;
#ifdef pgprot_writecombine
if (attrs & DMA_ATTR_WRITE_COMBINE)
return pgprot_writecombine(prot);
#endif
return pgprot_dmacoherent(prot);
}
But my worry is how this interacts with architectures that have an
uncached segment (mips, nios2, microblaze, extensa) where we'd have
the kernel access DMA_ATTR_WRITE_COMBINE mappigns using the uncached
segment, and userspace mmaps using pgprot_writecombine, which could
lead to aliasing issues. But then again mips already supports
DMA_ATTR_WRITE_COMBINE, so this must be ok somehow. I guess I'll
need to field that question to the relevant parties.
^ permalink raw reply
* Re: [PATCH] dma-mapping: fix page attributes for dma_mmap_*
From: Will Deacon @ 2019-08-01 16:44 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Shawn Anastasio, linuxppc-dev, Russell King, linux-kernel, iommu,
Catalin Marinas, Robin Murphy, linux-arm-kernel
In-Reply-To: <20190801163457.GB26588@lst.de>
On Thu, Aug 01, 2019 at 06:34:57PM +0200, Christoph Hellwig wrote:
> On Thu, Aug 01, 2019 at 05:23:06PM +0100, Will Deacon wrote:
> > > - if (!dev_is_dma_coherent(dev) || (attrs & DMA_ATTR_WRITE_COMBINE))
> > > - return pgprot_writecombine(prot);
> > > - return prot;
> > > + return pgprot_writecombine(prot);
> > > }
> >
> > Seems like a sensible cleanup to me:
> >
> > Acked-by: Will Deacon <will@kernel.org>
> >
> > Although arch_dma_mmap_pgprot() is a bit of a misnomer now that it only
> > gets involved in the non-coherent case.
>
> A better name is welcome.
How about arch_dma_noncoherent_mmap_pgprot() ? Too long?
> My other idea would be to just remove it entirely and do something like:
>
> #ifndef pgprot_dmacoherent
> #define pgprot_dmacoherent pgprot_noncached
> #endif
>
> pgprot_t dma_mmap_pgprot(struct device *dev, pgprot_t prot, unsigned long attrs)
> {
> if (dev_is_dma_coherent(dev) || (attrs & DMA_ATTR_NON_CONSISTENT))
> return prot;
> #ifdef pgprot_writecombine
> if (attrs & DMA_ATTR_WRITE_COMBINE)
> return pgprot_writecombine(prot);
> #endif
> return pgprot_dmacoherent(prot);
> }
Oh, I prefer that!
> But my worry is how this interacts with architectures that have an
> uncached segment (mips, nios2, microblaze, extensa) where we'd have
> the kernel access DMA_ATTR_WRITE_COMBINE mappigns using the uncached
> segment, and userspace mmaps using pgprot_writecombine, which could
> lead to aliasing issues. But then again mips already supports
> DMA_ATTR_WRITE_COMBINE, so this must be ok somehow. I guess I'll
> need to field that question to the relevant parties.
Or it's always been busted and happens to work out in practice...
Will
^ permalink raw reply
* Re: [PATCH] selftests/powerpc: Fix build failures with GCC 9
From: Segher Boessenkool @ 2019-08-01 18:24 UTC (permalink / raw)
To: Michael Ellerman; +Cc: linuxppc-dev
In-Reply-To: <20190801122628.25084-1-mpe@ellerman.id.au>
On Thu, Aug 01, 2019 at 10:26:28PM +1000, Michael Ellerman wrote:
> GCC 9 fails to build some of the ptrace TM tests, with errors such as:
>
> ptrace-tm-spd-vsx.c: In function 'tm_spd_vsx':
> ptrace-tm-spd-vsx.c:51:2: error: listing the stack pointer register 'r1' in a clobber list is deprecated [-Werror=deprecated]
> 51 | asm __volatile__(
> | ^~~
> ptrace-tm-spd-vsx.c:51:2: note: the value of the stack pointer after an 'asm' statement must be the same as it was before the statement
>
> Which is probably fair enough.
Maybe you shouldn't build the tests with -Werror though? (And you could
include the much more useful -Wextra while you're at it ;-) ).
> Some of these inline asm blocks are doing quite a lot and are probably
> pushing the boundaries of what's sane to do with inline asm,
These are just testcases, you sometimes need to do evil things there.
But yeah :-)
> but they shouldn't actually be returning with r1 modified.
But they *do* modify lr, and that one isnt't listed; I guess the r1
clobber was there because of the call in the asm, but that needs an
lr clobber, instead.
(Or it was because of the "or 1,1,1")?
Segher
^ permalink raw reply
* [PATCH v3 0/7] soc/fsl/qbman: Enable Kexec for DPAA1 devices
From: Roy Pledge @ 2019-08-01 20:16 UTC (permalink / raw)
To: linuxppc-dev@lists.ozlabs.org,
linux-arm-kernel@lists.infradead.org,
linux-kernel@vger.kernel.org, Leo Li
Cc: Roy Pledge, Laurentiu Tudor, Madalin-cristian Bucur
Most DPAA1 devices do not support a soft reset which is an issue if
Kexec starts a new kernel. This patch series allows Kexec to function
by detecting that the QBMan device was previously initialized.
The patches fix some issues with device cleanup as well as ensuring
that the location of the QBMan private memories has not changed
after the execution of the Kexec.
Changes since v1:
- Removed a bug fix and sent it separately to ease backporting
Changes since v2:
- Expliciitly flush FQD memory from cache on PPC before unmapping
Roy Pledge (7):
soc/fsl/qbman: Rework QBMan private memory setup
soc/fsl/qbman: Cleanup buffer pools if BMan was initialized prior to
bootup
soc/fsl/qbman: Cleanup QMan queues if device was already initialized
soc/fsl/qbman: Fix drain_mr_fqni()
soc/fsl/qbman: Disable interrupts during portal recovery
soc/fsl/qbman: Fixup qman_shutdown_fq()
soc/fsl/qbman: Update device tree with reserved memory
drivers/soc/fsl/qbman/bman.c | 17 ++++----
drivers/soc/fsl/qbman/bman_ccsr.c | 36 +++++++++++++++-
drivers/soc/fsl/qbman/bman_portal.c | 18 +++++++-
drivers/soc/fsl/qbman/bman_priv.h | 5 +++
drivers/soc/fsl/qbman/dpaa_sys.c | 63 ++++++++++++++++------------
drivers/soc/fsl/qbman/qman.c | 83 +++++++++++++++++++++++++++++--------
drivers/soc/fsl/qbman/qman_ccsr.c | 68 +++++++++++++++++++++++++++---
drivers/soc/fsl/qbman/qman_portal.c | 18 +++++++-
drivers/soc/fsl/qbman/qman_priv.h | 8 ++++
9 files changed, 255 insertions(+), 61 deletions(-)
--
2.7.4
^ permalink raw reply
* [PATCH v3 1/7] soc/fsl/qbman: Rework QBMan private memory setup
From: Roy Pledge @ 2019-08-01 20:16 UTC (permalink / raw)
To: linuxppc-dev@lists.ozlabs.org,
linux-arm-kernel@lists.infradead.org,
linux-kernel@vger.kernel.org, Leo Li
Cc: Roy Pledge, Laurentiu Tudor, Madalin-cristian Bucur
In-Reply-To: <1564690599-29713-1-git-send-email-roy.pledge@nxp.com>
Rework QBMan private memory setup so that the areas are not
zeroed if the device was previously initialized
If the QMan private memory was already initialized skip the PFDR
initialization.
Signed-off-by: Roy Pledge <roy.pledge@nxp.com>
---
drivers/soc/fsl/qbman/bman_ccsr.c | 26 +++++++++++++++++--
drivers/soc/fsl/qbman/dpaa_sys.c | 7 +++--
drivers/soc/fsl/qbman/qman_ccsr.c | 54 +++++++++++++++++++++++++++++++++++----
3 files changed, 76 insertions(+), 11 deletions(-)
diff --git a/drivers/soc/fsl/qbman/bman_ccsr.c b/drivers/soc/fsl/qbman/bman_ccsr.c
index 7c3cc96..dc6d7e5 100644
--- a/drivers/soc/fsl/qbman/bman_ccsr.c
+++ b/drivers/soc/fsl/qbman/bman_ccsr.c
@@ -97,17 +97,40 @@ static void bm_get_version(u16 *id, u8 *major, u8 *minor)
/* signal transactions for FBPRs with higher priority */
#define FBPR_AR_RPRIO_HI BIT(30)
-static void bm_set_memory(u64 ba, u32 size)
+/* Track if probe has occurred and if cleanup is required */
+static int __bman_probed;
+static int __bman_requires_cleanup;
+
+
+static int bm_set_memory(u64 ba, u32 size)
{
+ u32 bar, bare;
u32 exp = ilog2(size);
/* choke if size isn't within range */
DPAA_ASSERT(size >= 4096 && size <= 1024*1024*1024 &&
is_power_of_2(size));
/* choke if '[e]ba' has lower-alignment than 'size' */
DPAA_ASSERT(!(ba & (size - 1)));
+
+ /* Check to see if BMan has already been initialized */
+ bar = bm_ccsr_in(REG_FBPR_BAR);
+ if (bar) {
+ /* Maker sure ba == what was programmed) */
+ bare = bm_ccsr_in(REG_FBPR_BARE);
+ if (bare != upper_32_bits(ba) || bar != lower_32_bits(ba)) {
+ pr_err("Attempted to reinitialize BMan with different BAR, got 0x%llx read BARE=0x%x BAR=0x%x\n",
+ ba, bare, bar);
+ return -ENOMEM;
+ }
+ pr_info("BMan BAR already configured\n");
+ __bman_requires_cleanup = 1;
+ return 1;
+ }
+
bm_ccsr_out(REG_FBPR_BARE, upper_32_bits(ba));
bm_ccsr_out(REG_FBPR_BAR, lower_32_bits(ba));
bm_ccsr_out(REG_FBPR_AR, exp - 1);
+ return 0;
}
/*
@@ -120,7 +143,6 @@ static void bm_set_memory(u64 ba, u32 size)
*/
static dma_addr_t fbpr_a;
static size_t fbpr_sz;
-static int __bman_probed;
static int bman_fbpr(struct reserved_mem *rmem)
{
diff --git a/drivers/soc/fsl/qbman/dpaa_sys.c b/drivers/soc/fsl/qbman/dpaa_sys.c
index e6d48dc..3e0a7f3 100644
--- a/drivers/soc/fsl/qbman/dpaa_sys.c
+++ b/drivers/soc/fsl/qbman/dpaa_sys.c
@@ -40,6 +40,7 @@ int qbman_init_private_mem(struct device *dev, int idx, dma_addr_t *addr,
int ret;
struct device_node *mem_node;
u64 size64;
+ struct reserved_mem *rmem;
ret = of_reserved_mem_device_init_by_idx(dev, dev->of_node, idx);
if (ret) {
@@ -62,10 +63,8 @@ int qbman_init_private_mem(struct device *dev, int idx, dma_addr_t *addr,
return -ENODEV;
}
- if (!dma_alloc_coherent(dev, *size, addr, 0)) {
- dev_err(dev, "DMA Alloc memory failed\n");
- return -ENODEV;
- }
+ rmem = of_reserved_mem_lookup(mem_node);
+ *addr = rmem->base;
/*
* Disassociate the reserved memory area from the device
diff --git a/drivers/soc/fsl/qbman/qman_ccsr.c b/drivers/soc/fsl/qbman/qman_ccsr.c
index a6bb430..a3edefa 100644
--- a/drivers/soc/fsl/qbman/qman_ccsr.c
+++ b/drivers/soc/fsl/qbman/qman_ccsr.c
@@ -274,6 +274,7 @@ static u32 __iomem *qm_ccsr_start;
/* A SDQCR mask comprising all the available/visible pool channels */
static u32 qm_pools_sdqcr;
static int __qman_probed;
+static int __qman_requires_cleanup;
static inline u32 qm_ccsr_in(u32 offset)
{
@@ -340,19 +341,55 @@ static void qm_get_version(u16 *id, u8 *major, u8 *minor)
}
#define PFDR_AR_EN BIT(31)
-static void qm_set_memory(enum qm_memory memory, u64 ba, u32 size)
+static int qm_set_memory(enum qm_memory memory, u64 ba, u32 size)
{
+ void *ptr;
u32 offset = (memory == qm_memory_fqd) ? REG_FQD_BARE : REG_PFDR_BARE;
u32 exp = ilog2(size);
+ u32 bar, bare;
/* choke if size isn't within range */
DPAA_ASSERT((size >= 4096) && (size <= 1024*1024*1024) &&
is_power_of_2(size));
/* choke if 'ba' has lower-alignment than 'size' */
DPAA_ASSERT(!(ba & (size - 1)));
+
+ /* Check to see if QMan has already been initialized */
+ bar = qm_ccsr_in(offset + REG_offset_BAR);
+ if (bar) {
+ /* Maker sure ba == what was programmed) */
+ bare = qm_ccsr_in(offset);
+ if (bare != upper_32_bits(ba) || bar != lower_32_bits(ba)) {
+ pr_err("Attempted to reinitialize QMan with different BAR, got 0x%llx read BARE=0x%x BAR=0x%x\n",
+ ba, bare, bar);
+ return -ENOMEM;
+ }
+ __qman_requires_cleanup = 1;
+ /* Return 1 to indicate memory was previously programmed */
+ return 1;
+ }
+ /* Need to temporarily map the area to make sure it is zeroed */
+ ptr = memremap(ba, size, MEMREMAP_WB);
+ if (!ptr) {
+ pr_crit("memremap() of QMan private memory failed\n");
+ return -ENOMEM;
+ }
+ memset(ptr, 0, size);
+
+#ifdef CONFIG_PPC
+ /*
+ * PPC doesn't appear to flush the cache on memunmap() but the
+ * cache must be flushed since QMan does non coherent accesses
+ * to this memory
+ */
+ flush_dcache_range((unsigned long) ptr, (unsigned long) ptr+size);
+#endif
+ memunmap(ptr);
+
qm_ccsr_out(offset, upper_32_bits(ba));
qm_ccsr_out(offset + REG_offset_BAR, lower_32_bits(ba));
qm_ccsr_out(offset + REG_offset_AR, PFDR_AR_EN | (exp - 1));
+ return 0;
}
static void qm_set_pfdr_threshold(u32 th, u8 k)
@@ -571,12 +608,19 @@ static int qman_init_ccsr(struct device *dev)
int i, err;
/* FQD memory */
- qm_set_memory(qm_memory_fqd, fqd_a, fqd_sz);
+ err = qm_set_memory(qm_memory_fqd, fqd_a, fqd_sz);
+ if (err < 0)
+ return err;
/* PFDR memory */
- qm_set_memory(qm_memory_pfdr, pfdr_a, pfdr_sz);
- err = qm_init_pfdr(dev, 8, pfdr_sz / 64 - 8);
- if (err)
+ err = qm_set_memory(qm_memory_pfdr, pfdr_a, pfdr_sz);
+ if (err < 0)
return err;
+ /* Only initialize PFDRs if the QMan was not initialized before */
+ if (err == 0) {
+ err = qm_init_pfdr(dev, 8, pfdr_sz / 64 - 8);
+ if (err)
+ return err;
+ }
/* thresholds */
qm_set_pfdr_threshold(512, 64);
qm_set_sfdr_threshold(128);
--
2.7.4
^ permalink raw reply related
* [PATCH v3 2/7] soc/fsl/qbman: Cleanup buffer pools if BMan was initialized prior to bootup
From: Roy Pledge @ 2019-08-01 20:16 UTC (permalink / raw)
To: linuxppc-dev@lists.ozlabs.org,
linux-arm-kernel@lists.infradead.org,
linux-kernel@vger.kernel.org, Leo Li
Cc: Roy Pledge, Laurentiu Tudor, Madalin-cristian Bucur
In-Reply-To: <1564690599-29713-1-git-send-email-roy.pledge@nxp.com>
Clean the BMan buffer pools if the device had been initialized
previously. This will ensure a consistent state if the kernel
was soft restarted (kexec for example)
Signed-off-by: Roy Pledge <roy.pledge@nxp.com>
---
drivers/soc/fsl/qbman/bman.c | 17 +++++++++--------
drivers/soc/fsl/qbman/bman_ccsr.c | 10 ++++++++++
drivers/soc/fsl/qbman/bman_portal.c | 18 +++++++++++++++++-
drivers/soc/fsl/qbman/bman_priv.h | 5 +++++
4 files changed, 41 insertions(+), 9 deletions(-)
diff --git a/drivers/soc/fsl/qbman/bman.c b/drivers/soc/fsl/qbman/bman.c
index f84ab59..f4fb527 100644
--- a/drivers/soc/fsl/qbman/bman.c
+++ b/drivers/soc/fsl/qbman/bman.c
@@ -635,30 +635,31 @@ int bman_p_irqsource_add(struct bman_portal *p, u32 bits)
return 0;
}
-static int bm_shutdown_pool(u32 bpid)
+int bm_shutdown_pool(u32 bpid)
{
+ int err = 0;
struct bm_mc_command *bm_cmd;
union bm_mc_result *bm_res;
+
+ struct bman_portal *p = get_affine_portal();
while (1) {
- struct bman_portal *p = get_affine_portal();
/* Acquire buffers until empty */
bm_cmd = bm_mc_start(&p->p);
bm_cmd->bpid = bpid;
bm_mc_commit(&p->p, BM_MCC_VERB_CMD_ACQUIRE | 1);
if (!bm_mc_result_timeout(&p->p, &bm_res)) {
- put_affine_portal();
pr_crit("BMan Acquire Command timedout\n");
- return -ETIMEDOUT;
+ err = -ETIMEDOUT;
+ goto done;
}
if (!(bm_res->verb & BM_MCR_VERB_ACQUIRE_BUFCOUNT)) {
- put_affine_portal();
/* Pool is empty */
- return 0;
+ goto done;
}
- put_affine_portal();
}
-
+done:
+ put_affine_portal();
return 0;
}
diff --git a/drivers/soc/fsl/qbman/bman_ccsr.c b/drivers/soc/fsl/qbman/bman_ccsr.c
index dc6d7e5..cb24a08 100644
--- a/drivers/soc/fsl/qbman/bman_ccsr.c
+++ b/drivers/soc/fsl/qbman/bman_ccsr.c
@@ -195,6 +195,16 @@ int bman_is_probed(void)
}
EXPORT_SYMBOL_GPL(bman_is_probed);
+int bman_requires_cleanup(void)
+{
+ return __bman_requires_cleanup;
+}
+
+void bman_done_cleanup(void)
+{
+ __bman_requires_cleanup = 0;
+}
+
static int fsl_bman_probe(struct platform_device *pdev)
{
int ret, err_irq;
diff --git a/drivers/soc/fsl/qbman/bman_portal.c b/drivers/soc/fsl/qbman/bman_portal.c
index c78cc69..cc06d95 100644
--- a/drivers/soc/fsl/qbman/bman_portal.c
+++ b/drivers/soc/fsl/qbman/bman_portal.c
@@ -100,7 +100,7 @@ static int bman_portal_probe(struct platform_device *pdev)
struct device_node *node = dev->of_node;
struct bm_portal_config *pcfg;
struct resource *addr_phys[2];
- int irq, cpu, err;
+ int irq, cpu, err, i;
err = bman_is_probed();
if (!err)
@@ -181,6 +181,22 @@ static int bman_portal_probe(struct platform_device *pdev)
if (!cpu_online(cpu))
bman_offline_cpu(cpu);
+ if (__bman_portals_probed == 1 && bman_requires_cleanup()) {
+ /*
+ * BMan wasn't reset prior to boot (Kexec for example)
+ * Empty all the buffer pools so they are in reset state
+ */
+ for (i = 0; i < BM_POOL_MAX; i++) {
+ err = bm_shutdown_pool(i);
+ if (err) {
+ dev_err(dev, "Failed to shutdown bpool %d\n",
+ i);
+ goto err_portal_init;
+ }
+ }
+ bman_done_cleanup();
+ }
+
return 0;
err_portal_init:
diff --git a/drivers/soc/fsl/qbman/bman_priv.h b/drivers/soc/fsl/qbman/bman_priv.h
index 751ce90..aa3981e 100644
--- a/drivers/soc/fsl/qbman/bman_priv.h
+++ b/drivers/soc/fsl/qbman/bman_priv.h
@@ -76,3 +76,8 @@ int bman_p_irqsource_add(struct bman_portal *p, u32 bits);
const struct bm_portal_config *
bman_get_bm_portal_config(const struct bman_portal *portal);
+
+int bman_requires_cleanup(void);
+void bman_done_cleanup(void);
+
+int bm_shutdown_pool(u32 bpid);
--
2.7.4
^ permalink raw reply related
* [PATCH v3 3/7] soc/fsl/qbman: Cleanup QMan queues if device was already initialized
From: Roy Pledge @ 2019-08-01 20:16 UTC (permalink / raw)
To: linuxppc-dev@lists.ozlabs.org,
linux-arm-kernel@lists.infradead.org,
linux-kernel@vger.kernel.org, Leo Li
Cc: Roy Pledge, Laurentiu Tudor, Madalin-cristian Bucur
In-Reply-To: <1564690599-29713-1-git-send-email-roy.pledge@nxp.com>
If the QMan device was previously initialized make sure all the
frame queues are out of service once all the portals are probed.
This handles the case where the kernel is restarted without the
SoC being reset (kexec for example)
Signed-off-by: Roy Pledge <roy.pledge@nxp.com>
---
drivers/soc/fsl/qbman/qman.c | 4 ++--
drivers/soc/fsl/qbman/qman_ccsr.c | 13 ++++++++++++-
drivers/soc/fsl/qbman/qman_portal.c | 18 +++++++++++++++++-
drivers/soc/fsl/qbman/qman_priv.h | 7 +++++++
4 files changed, 38 insertions(+), 4 deletions(-)
diff --git a/drivers/soc/fsl/qbman/qman.c b/drivers/soc/fsl/qbman/qman.c
index 636f83f..f10f77d 100644
--- a/drivers/soc/fsl/qbman/qman.c
+++ b/drivers/soc/fsl/qbman/qman.c
@@ -2581,7 +2581,7 @@ static int _qm_dqrr_consume_and_match(struct qm_portal *p, u32 fqid, int s,
#define qm_dqrr_drain_nomatch(p) \
_qm_dqrr_consume_and_match(p, 0, 0, false)
-static int qman_shutdown_fq(u32 fqid)
+int qman_shutdown_fq(u32 fqid)
{
struct qman_portal *p;
struct device *dev;
@@ -2754,7 +2754,7 @@ static int qman_shutdown_fq(u32 fqid)
DPAA_ASSERT((mcr->verb & QM_MCR_VERB_MASK) ==
QM_MCR_VERB_ALTER_OOS);
- if (mcr->result) {
+ if (mcr->result != QM_MCR_RESULT_OK) {
dev_err(dev, "OOS fail: FQ 0x%x (0x%x)\n",
fqid, mcr->result);
ret = -EIO;
diff --git a/drivers/soc/fsl/qbman/qman_ccsr.c b/drivers/soc/fsl/qbman/qman_ccsr.c
index a3edefa..709661b7b 100644
--- a/drivers/soc/fsl/qbman/qman_ccsr.c
+++ b/drivers/soc/fsl/qbman/qman_ccsr.c
@@ -492,7 +492,7 @@ RESERVEDMEM_OF_DECLARE(qman_pfdr, "fsl,qman-pfdr", qman_pfdr);
#endif
-static unsigned int qm_get_fqid_maxcnt(void)
+unsigned int qm_get_fqid_maxcnt(void)
{
return fqd_sz / 64;
}
@@ -737,6 +737,17 @@ int qman_is_probed(void)
}
EXPORT_SYMBOL_GPL(qman_is_probed);
+int qman_requires_cleanup(void)
+{
+ return __qman_requires_cleanup;
+}
+
+void qman_done_cleanup(void)
+{
+ __qman_requires_cleanup = 0;
+}
+
+
static int fsl_qman_probe(struct platform_device *pdev)
{
struct device *dev = &pdev->dev;
diff --git a/drivers/soc/fsl/qbman/qman_portal.c b/drivers/soc/fsl/qbman/qman_portal.c
index 75717bc..153727c 100644
--- a/drivers/soc/fsl/qbman/qman_portal.c
+++ b/drivers/soc/fsl/qbman/qman_portal.c
@@ -233,7 +233,7 @@ static int qman_portal_probe(struct platform_device *pdev)
struct device_node *node = dev->of_node;
struct qm_portal_config *pcfg;
struct resource *addr_phys[2];
- int irq, cpu, err;
+ int irq, cpu, err, i;
u32 val;
err = qman_is_probed();
@@ -328,6 +328,22 @@ static int qman_portal_probe(struct platform_device *pdev)
if (!cpu_online(cpu))
qman_offline_cpu(cpu);
+ if (__qman_portals_probed == 1 && qman_requires_cleanup()) {
+ /*
+ * QMan wasn't reset prior to boot (Kexec for example)
+ * Empty all the frame queues so they are in reset state
+ */
+ for (i = 0; i < qm_get_fqid_maxcnt(); i++) {
+ err = qman_shutdown_fq(i);
+ if (err) {
+ dev_err(dev, "Failed to shutdown frame queue %d\n",
+ i);
+ goto err_portal_init;
+ }
+ }
+ qman_done_cleanup();
+ }
+
return 0;
err_portal_init:
diff --git a/drivers/soc/fsl/qbman/qman_priv.h b/drivers/soc/fsl/qbman/qman_priv.h
index 0451571..a8a35fe 100644
--- a/drivers/soc/fsl/qbman/qman_priv.h
+++ b/drivers/soc/fsl/qbman/qman_priv.h
@@ -272,3 +272,10 @@ extern struct qman_portal *affine_portals[NR_CPUS];
extern struct qman_portal *qman_dma_portal;
const struct qm_portal_config *qman_get_qm_portal_config(
struct qman_portal *portal);
+
+unsigned int qm_get_fqid_maxcnt(void);
+
+int qman_shutdown_fq(u32 fqid);
+
+int qman_requires_cleanup(void);
+void qman_done_cleanup(void);
--
2.7.4
^ permalink raw reply related
* [PATCH v3 4/7] soc/fsl/qbman: Fix drain_mr_fqni()
From: Roy Pledge @ 2019-08-01 20:16 UTC (permalink / raw)
To: linuxppc-dev@lists.ozlabs.org,
linux-arm-kernel@lists.infradead.org,
linux-kernel@vger.kernel.org, Leo Li
Cc: Roy Pledge, Laurentiu Tudor, Madalin-cristian Bucur
In-Reply-To: <1564690599-29713-1-git-send-email-roy.pledge@nxp.com>
The drain_mr_fqni() function may be called fron uninterruptable
context so convert the msleep() to an mdelay(). Also ensure that
the valid bit is updated while polling.
Signed-off-by: Roy Pledge <roy.pledge@nxp.com>
---
drivers/soc/fsl/qbman/qman.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/drivers/soc/fsl/qbman/qman.c b/drivers/soc/fsl/qbman/qman.c
index f10f77d..2989504 100644
--- a/drivers/soc/fsl/qbman/qman.c
+++ b/drivers/soc/fsl/qbman/qman.c
@@ -1164,6 +1164,7 @@ static int drain_mr_fqrni(struct qm_portal *p)
{
const union qm_mr_entry *msg;
loop:
+ qm_mr_pvb_update(p);
msg = qm_mr_current(p);
if (!msg) {
/*
@@ -1180,7 +1181,8 @@ static int drain_mr_fqrni(struct qm_portal *p)
* entries well before the ring has been fully consumed, so
* we're being *really* paranoid here.
*/
- msleep(1);
+ mdelay(1);
+ qm_mr_pvb_update(p);
msg = qm_mr_current(p);
if (!msg)
return 0;
--
2.7.4
^ permalink raw reply related
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