* Re: [PATCH] KVM: PPC: Book3S HV: Fix comment typos
From: Ritesh Harjani @ 2026-07-09 7:01 UTC (permalink / raw)
To: Salman Farooq, maddy
Cc: npiggin, mpe, chleroy, linuxppc-dev, kvm, linux-kernel,
Salman Farooq
In-Reply-To: <20260702030344.2211462-1-salman.farooq3310@gmail.com>
Salman Farooq <salman.farooq3310@gmail.com> writes:
> Fix two instances of a misspelled word in comments.
>
Not sure what this will really get us and what is our maintainer take on
such spelling mistake fixes. I definitely hope we won't churn out a lot
of patches just to fix simple spell mistakes.
But FWIW, the change looks good to me, so:
Reviewed-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
> Signed-off-by: Salman Farooq <salman.farooq3310@gmail.com>
> ---
> arch/powerpc/kvm/book3s_hv_uvmem.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/arch/powerpc/kvm/book3s_hv_uvmem.c b/arch/powerpc/kvm/book3s_hv_uvmem.c
> index 5fbb95d90e99..6ac8364a7639 100644
> --- a/arch/powerpc/kvm/book3s_hv_uvmem.c
> +++ b/arch/powerpc/kvm/book3s_hv_uvmem.c
> @@ -163,7 +163,7 @@ static DEFINE_SPINLOCK(kvmppc_uvmem_bitmap_lock);
> * All its GFNs are moved to Normal-GFNs.
> *
> * UV_TERMINATE transitions the secure-VM back to normal-VM. All
> - * the secure-GFN and shared-GFNs are tranistioned to normal-GFN
> + * the secure-GFN and shared-GFNs are transitioned to normal-GFN
> * Note: The contents of the normal-GFN is undefined at this point.
> *
> * GFN state implementation:
> @@ -1010,7 +1010,7 @@ static vm_fault_t kvmppc_uvmem_migrate_to_ram(struct vm_fault *vmf)
> /*
> * Release the device PFN back to the pool
> *
> - * Gets called when secure GFN tranistions from a secure-PFN
> + * Gets called when secure GFN transitions from a secure-PFN
> * to a normal PFN during H_SVM_PAGE_OUT.
> * Gets called with kvm->arch.uvmem_lock held.
> */
> --
> 2.55.0
^ permalink raw reply
* Re: [PATCH 3/3] powerpc/boot: Fix treeboot-akebono CPU node lookup check
From: Ritesh Harjani @ 2026-07-09 6:52 UTC (permalink / raw)
To: Thorsten Blum, Madhavan Srinivasan, Michael Ellerman,
Nicholas Piggin, Christophe Leroy (CS GROUP), Josh Boyer,
Grant Likely, Tony Breeds, Alistair Popple,
Benjamin Herrenschmidt
Cc: linuxppc-dev, linux-kernel, Thorsten Blum
In-Reply-To: <20260702211554.56923-6-thorsten.blum@linux.dev>
Thorsten Blum <thorsten.blum@linux.dev> writes:
> fdt_node_offset_by_prop_value() returns a negative error code on
> failure - fix the check accordingly.
>
> Fixes: 2a2c74b2efcb ("IBM Akebono: Add the Akebono platform")
> Signed-off-by: Thorsten Blum <thorsten.blum@linux.dev>
Looks good to me. Please feel free to add:
Reviewed-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
> ---
> arch/powerpc/boot/treeboot-akebono.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/powerpc/boot/treeboot-akebono.c b/arch/powerpc/boot/treeboot-akebono.c
> index e3cc2599869c..1b529037480f 100644
> --- a/arch/powerpc/boot/treeboot-akebono.c
> +++ b/arch/powerpc/boot/treeboot-akebono.c
> @@ -146,7 +146,7 @@ void platform_init(char *userdata)
>
> node = fdt_node_offset_by_prop_value(_dtb_start, -1, "device_type",
> "cpu", sizeof("cpu"));
> - if (!node)
> + if (node < 0)
> fatal("Cannot find cpu node\n");
> timebase = fdt_getprop(_dtb_start, node, "timebase-frequency", &size);
> if (timebase && (size == 4))
^ permalink raw reply
* Re: [PATCH 2/3] powerpc/boot: Fix treeboot-currituck CPU node lookup check
From: Ritesh Harjani @ 2026-07-09 6:52 UTC (permalink / raw)
To: Thorsten Blum, Madhavan Srinivasan, Michael Ellerman,
Nicholas Piggin, Christophe Leroy (CS GROUP), Josh Boyer,
Grant Likely, Tony Breeds, Alistair Popple,
Benjamin Herrenschmidt
Cc: linuxppc-dev, linux-kernel, Thorsten Blum
In-Reply-To: <20260702211554.56923-5-thorsten.blum@linux.dev>
Thorsten Blum <thorsten.blum@linux.dev> writes:
> fdt_node_offset_by_prop_value() returns a negative error code on
> failure - fix the check accordingly.
>
> Fixes: 228d55053397 ("powerpc/47x: Add support for the new IBM currituck platform")
> Signed-off-by: Thorsten Blum <thorsten.blum@linux.dev>
Looks good to me. Please feel free to add:
Reviewed-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
> ---
> arch/powerpc/boot/treeboot-currituck.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/powerpc/boot/treeboot-currituck.c b/arch/powerpc/boot/treeboot-currituck.c
> index d53e8a592f81..5b5363b74f9f 100644
> --- a/arch/powerpc/boot/treeboot-currituck.c
> +++ b/arch/powerpc/boot/treeboot-currituck.c
> @@ -102,7 +102,7 @@ void platform_init(void)
>
> node = fdt_node_offset_by_prop_value(_dtb_start, -1, "device_type",
> "cpu", sizeof("cpu"));
> - if (!node)
> + if (node < 0)
> fatal("Cannot find cpu node\n");
> timebase = fdt_getprop(_dtb_start, node, "timebase-frequency", &size);
> if (timebase && (size == 4))
^ permalink raw reply
* Re: [PATCH 1/3] powerpc/boot: Fix simpleboot CPU node lookup check
From: Ritesh Harjani @ 2026-07-09 6:26 UTC (permalink / raw)
To: Thorsten Blum, Madhavan Srinivasan, Michael Ellerman,
Nicholas Piggin, Christophe Leroy (CS GROUP), Josh Boyer,
Grant Likely, Tony Breeds, Alistair Popple,
Benjamin Herrenschmidt
Cc: linuxppc-dev, linux-kernel, Thorsten Blum
In-Reply-To: <20260702211554.56923-4-thorsten.blum@linux.dev>
Thorsten Blum <thorsten.blum@linux.dev> writes:
> fdt_node_offset_by_prop_value() returns a negative error code on
> failure - fix the check accordingly.
>
> Fixes: d2477b5cc8ca ("[POWERPC] bootwrapper: Add a firmware-independent simpleboot target.")
> Signed-off-by: Thorsten Blum <thorsten.blum@linux.dev>
yup, if it cannot find the node, then it should return a negative error
code, since cpu node is never the root node. The same thing was anyway
properly done for memory device node in the same function few lines
before...
/* Find the memory range */
node = fdt_node_offset_by_prop_value(_dtb_start, -1, "device_type",
"memory", sizeof("memory"));
if (node < 0)
fatal("Cannot find memory node\n");
...but for cpu node, it uses a wrong comparison check.
Looks good to me. Please feel free to add:
Reviewed-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
> ---
> arch/powerpc/boot/simpleboot.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/powerpc/boot/simpleboot.c b/arch/powerpc/boot/simpleboot.c
> index c80691d83880..27591df41e9e 100644
> --- a/arch/powerpc/boot/simpleboot.c
> +++ b/arch/powerpc/boot/simpleboot.c
> @@ -68,7 +68,7 @@ void platform_init(unsigned long r3, unsigned long r4, unsigned long r5,
> /* finally, setup the timebase */
> node = fdt_node_offset_by_prop_value(_dtb_start, -1, "device_type",
> "cpu", sizeof("cpu"));
> - if (!node)
> + if (node < 0)
> fatal("Cannot find cpu node\n");
> timebase = fdt_getprop(_dtb_start, node, "timebase-frequency", &size);
> if (timebase && (size == 4))
^ permalink raw reply
* Re: [kvm-unit-tests PATCH 4/6] powerpc: gitlab CI update
From: Thomas Huth @ 2026-07-09 6:46 UTC (permalink / raw)
To: Chinmay Rath
Cc: npiggin, harshpb, lvivier, linuxppc-dev, kvm, andrew.jones, sbhat
In-Reply-To: <20260708105122.1177210-5-rathc@linux.ibm.com>
On 08/07/2026 12.51, Chinmay Rath wrote:
> From: Nicholas Piggin <npiggin@gmail.com>
>
> Adds a few additional tests that are known to work in CI.
>
> To control overhead, ppc64be is used to test 64k page size, ppc64le is
> used to test 4k page size.
>
> Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
> Signed-off-by: Chinmay Rath <rathc@linux.ibm.com>
> ---
> .gitlab-ci.yml | 19 ++++++++++++++-----
> powerpc/unittests.cfg | 9 ++++++++-
> 2 files changed, 22 insertions(+), 6 deletions(-)
>
> diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
> index ab1be5f4..b28d1eb5 100644
> --- a/.gitlab-ci.yml
> +++ b/.gitlab-ci.yml
> @@ -190,18 +190,21 @@ build-ppc64be:
> - dnf install -y qemu-system-ppc gcc-powerpc64-linux-gnu nmap-ncat
> - mkdir build
> - cd build
> - - ../configure --arch=ppc64 --endian=big --cross-prefix=powerpc64-linux-gnu-
> + - ../configure --arch=ppc64 --endian=big --page-size=64k --cross-prefix=powerpc64-linux-gnu-
> - make -j2
> - - ACCEL=tcg ./run_tests.sh
> + - ACCEL=tcg MAX_SMP=8 ./run_tests.sh
> atomics
> + atomics-migration
> emulator
> interrupts
> mmu
> + pmu
> rtas-get-time-of-day
> rtas-get-time-of-day-base
> rtas-set-time-of-day
> selftest-migration
> selftest-migration-skip
> + selftest-migration-ci
> selftest-setup
> sieve
> smp
> @@ -209,10 +212,12 @@ build-ppc64be:
> spapr_hcall
> spapr_vpa
> sprs
> + sprs-migration
> timebase
> timebase-icount
> | tee results.txt
> - - if grep -q FAIL results.txt ; then exit 1 ; fi
> + - grep -q PASS results.txt && ! grep -q FAIL results.txt
> +
Remove the superfluous empty line, please. We don't do this in other spots
between the jobs in this file, so let's stay consistent here, too.
>
> build-ppc64le:
> extends: .intree_template
> @@ -220,16 +225,19 @@ build-ppc64le:
> - dnf install -y qemu-system-ppc gcc-powerpc64-linux-gnu nmap-ncat
> - ./configure --arch=ppc64 --endian=little --cross-prefix=powerpc64-linux-gnu-
> - make -j2
> - - ACCEL=tcg ./run_tests.sh
> + - ACCEL=tcg MAX_SMP=8 ./run_tests.sh
> atomics
> + atomics-migration
> emulator
> interrupts
> mmu
> + pmu
> rtas-get-time-of-day
> rtas-get-time-of-day-base
> rtas-set-time-of-day
> selftest-migration
> selftest-migration-skip
> + selftest-migration-ci
> selftest-setup
> sieve
> smp
> @@ -237,10 +245,11 @@ build-ppc64le:
> spapr_hcall
> spapr_vpa
> sprs
> + sprs-migration
> timebase
> timebase-icount
> | tee results.txt
> - - if grep -q FAIL results.txt ; then exit 1 ; fi
> + - grep -q PASS results.txt && ! grep -q FAIL results.txt
>
> build-riscv32:
> extends: .outoftree_template
> diff --git a/powerpc/unittests.cfg b/powerpc/unittests.cfg
> index 60c73086..62630ead 100644
> --- a/powerpc/unittests.cfg
> +++ b/powerpc/unittests.cfg
> @@ -24,6 +24,14 @@ file = selftest-migration.elf
> machine = pseries
> groups = selftest migration
>
> +# make a kvm-only version for CI for now
> +# TCG accel support with Qemu and distro release used in gitlab CI needs to checked.
> +[selftest-migration-ci]
> +file = selftest-migration.elf
> +machine = pseries
> +groups = nodefault selftest migration
> +accel = kvm
This does not make sense (or I don't understand it) - gitlab-CI currently
only runs with TCG for the ppc64 jobs, so adding a separate test for this
with "accel = kvm" just adds a job there that will always be skipped. What
should this be good for?
> [selftest-migration-skip]
> file = selftest-migration.elf
> machine = pseries
> @@ -94,7 +102,6 @@ accel = tcg,thread=single
>
> [atomics]
> file = atomics.elf
> -smp = 2
Why are you removing the smp = 2 here? As far as I can see, the atomics test
uses multiple CPUs, so this looks wrong?
Thomas
^ permalink raw reply
* Re: [PATCH v16 18/18] arm64: vdso: Expose sigreturn address on vdso to the kernel
From: Jinjie Ruan @ 2026-07-09 6:37 UTC (permalink / raw)
To: Thomas Weißschuh
Cc: oleg, richard.henderson, mattst88, linmag7, linux,
catalin.marinas, will, kees, guoren, chenhuacai, kernel, geert,
tsbogend, James.Bottomley, deller, maddy, mpe, npiggin, chleroy,
pjw, palmer, aou, alex, hca, gor, agordeev, borntraeger, svens,
ysato, dalias, glaubitz, richard, anton.ivanov, johannes, luto,
tglx, mingo, bp, dave.hansen, hpa, chris, jcmvbkbc, peterz, wad,
thuth, mark.rutland, ada.coupriediaz, kevin.brodsky, linusw,
yeoreum.yun, song, james.morse, anshuman.khandual, broonie,
liqiang01, pengcan, ryan.roberts, yangtiezhu, sshegde, mchauras,
austin.kim, jchrist, arnd, sohil.mehta, andrew.cooper3, jgross,
kas, x86, linux-alpha, linux-kernel, linux-arm-kernel, linux-mm,
linux-csky, loongarch, linux-m68k, linux-mips, linux-parisc,
linuxppc-dev, linux-riscv, linux-s390, linux-sh, linux-um
In-Reply-To: <20260630172619-6c78c2e2-c2d3-44be-88e6-6c82de12baf4@linutronix.de>
On 6/30/2026 11:32 PM, Thomas Weißschuh wrote:
> On Mon, Jun 29, 2026 at 09:06:16PM +0800, Jinjie Ruan wrote:
>> Syscall User Dispatch (SUD) requires the signal trampoline code to not be
>> intercepted. This is necessary to support returning with a locked selector
>> while avoiding infinite recursion back into the signal handler.
>>
>> Implement arch_syscall_is_vdso_sigreturn() for arm64 to support this
>> exclusion mechanism. For native 64-bit tasks, it checks whether the current
>> PC matches the 'svc #0' instruction inside the vDSO sigreturn trampoline.
>>
>> SYM_CODE_START(__kernel_rt_sigreturn)
>> mov x8, #__NR_rt_sigreturn
>> svc #0
>> SYM_CODE_END(__kernel_rt_sigreturn)
>>
>> For COMPAT tasks, it verifies if the instruction falls within
>> the architecture's 'sigpage' range, allowing the kernel to safely bypass
>> dispatching syscalls originating from these areas back to userspace.
>>
>> Suggested-by: Kevin Brodsky <kevin.brodsky@arm.com>
>> Suggested-by: kemal <kmal@cock.li>
>> Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
>> ---
>> arch/arm64/include/asm/elf.h | 1 +
>> arch/arm64/kernel/vdso.c | 16 ++++++++++++++++
>> 2 files changed, 17 insertions(+)
>>
>> diff --git a/arch/arm64/include/asm/elf.h b/arch/arm64/include/asm/elf.h
>> index d2779d604c7b..f43d4dea916a 100644
>> --- a/arch/arm64/include/asm/elf.h
>> +++ b/arch/arm64/include/asm/elf.h
>> @@ -185,6 +185,7 @@ do { \
>> struct linux_binprm;
>> extern int arch_setup_additional_pages(struct linux_binprm *bprm,
>> int uses_interp);
>> +extern bool arch_syscall_is_vdso_sigreturn(struct pt_regs *regs);
>
> A header for that?
Sure, it need a cleanup, as currently, there is no syscall_user_dispatch
header file related to the architecture.
>
>> /* 1GB of VA */
>> #ifdef CONFIG_COMPAT
>> diff --git a/arch/arm64/kernel/vdso.c b/arch/arm64/kernel/vdso.c
>> index 592dd8668de4..5a0314a3c26e 100644
>> --- a/arch/arm64/kernel/vdso.c
>> +++ b/arch/arm64/kernel/vdso.c
>> @@ -343,3 +343,19 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
>>
>> return ret;
>> }
>> +
>> +bool arch_syscall_is_vdso_sigreturn(struct pt_regs *regs)
>> +{
>> + unsigned long sigtramp;
>> +
>> +#ifdef CONFIG_COMPAT
>> + if (is_compat_task()) {
>> + unsigned long sigpage = (unsigned long)current->mm->context.sigpage;
>> +
>> + return regs->pc >= sigpage && regs->pc < (sigpage + PAGE_SIZE);
>> + }
>> +#endif
>> + sigtramp = (unsigned long)VDSO_SYMBOL(current->mm->context.vdso, sigtramp);
>> +
>> + return regs->pc == (sigtramp + 8);
>
> Instead of hardcoding 'sigtramp + 8' you could add a new label to the 'svc #0'
> instruction and use that with VDSO_SYMBOL().
It seems that the modification of __kernel_rt_sigreturn() is not recommended
>
>
> Thomas
^ permalink raw reply
* [PATCH v3 2/2] powerpc: enable to run posix cpu timers in task context
From: Vishal Chourasia @ 2026-07-09 6:29 UTC (permalink / raw)
To: maddy
Cc: peterz, frederic, npiggin, mpe, chleroy, sshegde, amachhiw,
vaibhav, harshpb, gautam, linuxppc-dev, kvm, linux-kernel,
Vishal Chourasia
In-Reply-To: <20260709062940.1679427-2-vishalc@linux.ibm.com>
From: Shrikanth Hegde <sshegde@linux.ibm.com>
From: Shrikanth Hegde <sshegde@linux.ibm.com>
Now that all kvm entry to guest paths handle the task work
using the generic framework, enable HAVE_POSIX_CPU_TIMERS_TASK_WORK
which allows running posix cpu timers in task context instead of running
them in hardirq. This would is a necessary step towards enabling
PREEMPT_RT on powerNV systems.
Signed-off-by: Shrikanth Hegde <sshegde@linux.ibm.com>
Signed-off-by: Vishal Chourasia <vishalc@linux.ibm.com>
---
arch/powerpc/Kconfig | 1 +
1 file changed, 1 insertion(+)
diff --git a/arch/powerpc/Kconfig b/arch/powerpc/Kconfig
index f7ce5fff81f0..51555a0b1a26 100644
--- a/arch/powerpc/Kconfig
+++ b/arch/powerpc/Kconfig
@@ -281,6 +281,7 @@ config PPC
select HAVE_PERF_REGS
select HAVE_PERF_USER_STACK_DUMP
select HAVE_PREEMPT_DYNAMIC_KEY
+ select HAVE_POSIX_CPU_TIMERS_TASK_WORK
select HAVE_RETHOOK if KPROBES
select HAVE_REGS_AND_STACK_ACCESS_API
select HAVE_RELIABLE_STACKTRACE
--
2.54.0
^ permalink raw reply related
* [PATCH v3 1/2] KVM: powerpc: Use generic xfer to guest work function
From: Vishal Chourasia @ 2026-07-09 6:29 UTC (permalink / raw)
To: maddy
Cc: peterz, frederic, npiggin, mpe, chleroy, sshegde, amachhiw,
vaibhav, harshpb, gautam, linuxppc-dev, kvm, linux-kernel,
Vishal Chourasia
In-Reply-To: <20260709062940.1679427-2-vishalc@linux.ibm.com>
Since commit 2cd571245b43 ("sched/fair: Add related data structure for
task based throttle") in v6.18, CFS bandwidth throttling no longer
dequeues a task directly; it queues task_work via TWA_RESUME and sets
TIF_NOTIFY_RESUME, relying on that work running before the task returns
to guest/user mode. The powerpc KVM run loops only checked for reschedule
and signals, never TIF_NOTIFY_RESUME, so the deferred throttle never ran
while a vCPU stayed in the run loop: a CPU-bound guest that rarely exits
to userspace ran far past its cpu.max quota and then appeared frozen for
minutes while the accrued throttle debt was repaid.
Use the generic infrastructure to check for and handle pending work
before transitioning into guest mode, replacing the open-coded
need_resched() and cond_resched() checks in the Book3S HV run loops and
in the common kvmppc_prepare_to_enter() used by the Book3S PR and BookE
run loops. The redundant signal_pending() recheck (and its sigpend label)
in kvmhv_run_single_vcpu() is also dropped, as
xfer_to_guest_mode_work_pending() is a superset of it.
This picks up handling for TIF_NOTIFY_RESUME, which was previously
ignored, meaning task work will now be correctly handled on every
guest re-entry.
Selecting VIRT_XFER_TO_GUEST_WORK disables RCU's last-resort self-IPI
fallback for vCPU tasks (see rcu_irq_work_resched()), which on
nohz_full CPUs was what forced a reschedule for deferred rcuog wakeups
queued right before guest entry. Take over that obligation the same way
x86 and s390 do: call xfer_to_guest_mode_prepare() with IRQs disabled
immediately before the final xfer_to_guest_mode_work_pending() check at
each guest-entry gate (kvmhv_run_single_vcpu(), kvmppc_run_core() and
kvmppc_prepare_to_enter()).
In kvmppc_prepare_to_enter(), IRQs are now disabled with
local_irq_disable() before hard_irq_disable(): on 32-bit,
hard_irq_disable() is a raw MSR[EE] clear that bypasses the
lockdep/irq-tracing state, and the strict xfer_to_guest_mode helpers
assert that IRQs are seen as disabled. This also allows upgrading the
racy __xfer_to_guest_mode_work_pending() check to the asserting
variant, as this loop is the terminal gate for the PR and BookE paths.
In kvmhv_run_single_vcpu(), the -EINTR exit and the pre-existing
kvmhv_setup_mmu() failure exit now leave via the done label instead of
returning directly, keeping the run_vcpu enter/exit tracepoints
balanced and vcpu->arch.ret consistent with the returned value.
In kvmppc_prepare_to_enter() the generic helper accounts the signal exit
(vcpu->stat.signal_exits and KVM_EXIT_INTR) but does not set the exit
type, so kvmppc_set_exit_type(SIGNAL_EXITS) is retained on the signal
path to preserve the E500 CONFIG_KVM_EXIT_TIMING histogram; it is a no-op
otherwise.
Signed-off-by: Vishal Chourasia <vishalc@linux.ibm.com>
---
arch/powerpc/kvm/Kconfig | 1 +
arch/powerpc/kvm/book3s_hv.c | 39 +++++++++++++++++++++++-------------
arch/powerpc/kvm/powerpc.c | 37 +++++++++++++++++++++++++---------
3 files changed, 54 insertions(+), 23 deletions(-)
diff --git a/arch/powerpc/kvm/Kconfig b/arch/powerpc/kvm/Kconfig
index 9a0d1c1aca6c..b6bc2fc86dca 100644
--- a/arch/powerpc/kvm/Kconfig
+++ b/arch/powerpc/kvm/Kconfig
@@ -22,6 +22,7 @@ config KVM
select KVM_COMMON
select KVM_VFIO
select HAVE_KVM_IRQ_BYPASS
+ select VIRT_XFER_TO_GUEST_WORK
config KVM_BOOK3S_HANDLER
bool
diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 61dbeea317f3..3cfe9a7be9c6 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -3853,7 +3853,8 @@ static noinline void kvmppc_run_core(struct kvmppc_vcore *vc)
*/
local_irq_disable();
hard_irq_disable();
- if (lazy_irq_pending() || need_resched() ||
+ xfer_to_guest_mode_prepare();
+ if (lazy_irq_pending() || xfer_to_guest_mode_work_pending() ||
recheck_signals_and_mmu(&core_info)) {
local_irq_enable();
vc->vcore_state = VCORE_INACTIVE;
@@ -4824,10 +4825,16 @@ static int kvmppc_run_vcpu(struct kvm_vcpu *vcpu)
vc->runner = vcpu;
if (n_ceded == vc->n_runnable) {
kvmppc_vcore_blocked(vc);
- } else if (need_resched()) {
+ } else if (__xfer_to_guest_mode_work_pending()) {
kvmppc_vcore_preempt(vc);
- /* Let something else run */
- cond_resched_lock(&vc->lock);
+ /*
+ * Let something else run. The raw helper is used as
+ * signal exits are accounted by this path already;
+ * it may schedule(), so drop the vcore lock.
+ */
+ spin_unlock(&vc->lock);
+ xfer_to_guest_mode_handle_work();
+ spin_lock(&vc->lock);
if (vc->vcore_state == VCORE_PREEMPT)
kvmppc_vcore_end_preempt(vc);
} else {
@@ -4895,12 +4902,16 @@ int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
run->exit_reason = KVM_EXIT_FAIL_ENTRY;
run->fail_entry.hardware_entry_failure_reason = 0;
vcpu->arch.ret = r;
- return r;
+ goto done;
}
}
- if (need_resched())
- cond_resched();
+ r = kvm_xfer_to_guest_mode_handle_work(vcpu);
+ if (r) {
+ /* -EINTR: signal pending, exit to userspace (KVM_EXIT_INTR) */
+ vcpu->arch.ret = r;
+ goto done;
+ }
kvmppc_update_vpas(vcpu);
@@ -4914,9 +4925,13 @@ int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
vcpu->arch.state = KVMPPC_VCPU_RUNNABLE;
- if (signal_pending(current))
- goto sigpend;
- if (need_resched() || !kvm->arch.mmu_ready)
+ xfer_to_guest_mode_prepare();
+
+ /*
+ * IRQs are disabled here, so on pending work bail to the outer loop,
+ * which handles it via kvm_xfer_to_guest_mode_handle_work() above.
+ */
+ if (xfer_to_guest_mode_work_pending() || !kvm->arch.mmu_ready)
goto out;
vcpu->cpu = pcpu;
@@ -5068,10 +5083,6 @@ int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
return vcpu->arch.ret;
- sigpend:
- vcpu->stat.signal_exits++;
- run->exit_reason = KVM_EXIT_INTR;
- vcpu->arch.ret = -EINTR;
out:
vcpu->cpu = -1;
vcpu->arch.thread_cpu = -1;
diff --git a/arch/powerpc/kvm/powerpc.c b/arch/powerpc/kvm/powerpc.c
index 00302399fc37..be5e48ae0c6c 100644
--- a/arch/powerpc/kvm/powerpc.c
+++ b/arch/powerpc/kvm/powerpc.c
@@ -81,23 +81,41 @@ int kvmppc_prepare_to_enter(struct kvm_vcpu *vcpu)
int r;
WARN_ON(irqs_disabled());
+ /*
+ * local_irq_disable() first: on 32-bit, hard_irq_disable() alone is a
+ * raw MSR[EE] clear that bypasses the lockdep/irq-tracing state, and
+ * the xfer_to_guest_mode helpers assert IRQs are seen as disabled.
+ */
+ local_irq_disable();
hard_irq_disable();
while (true) {
- if (need_resched()) {
+ xfer_to_guest_mode_prepare();
+
+ if (xfer_to_guest_mode_work_pending()) {
+ /*
+ * The helper must run with IRQs enabled and may
+ * schedule(). On a pending signal it returns -EINTR
+ * with run->exit_reason and vcpu->stat.signal_exits
+ * already set, so just return to userspace.
+ */
local_irq_enable();
- cond_resched();
+ r = kvm_xfer_to_guest_mode_handle_work(vcpu);
+ local_irq_disable();
hard_irq_disable();
+ if (r) {
+ /*
+ * The generic helper does not set the exit
+ * type; record it for the E500
+ * CONFIG_KVM_EXIT_TIMING histogram (a no-op
+ * otherwise).
+ */
+ kvmppc_set_exit_type(vcpu, SIGNAL_EXITS);
+ break;
+ }
continue;
}
- if (signal_pending(current)) {
- kvmppc_account_exit(vcpu, SIGNAL_EXITS);
- vcpu->run->exit_reason = KVM_EXIT_INTR;
- r = -EINTR;
- break;
- }
-
vcpu->mode = IN_GUEST_MODE;
/*
@@ -116,6 +134,7 @@ int kvmppc_prepare_to_enter(struct kvm_vcpu *vcpu)
local_irq_enable();
trace_kvm_check_requests(vcpu);
r = kvmppc_core_check_requests(vcpu);
+ local_irq_disable();
hard_irq_disable();
if (r > 0)
continue;
--
2.54.0
^ permalink raw reply related
* [PATCH v3 0/2] KVM: powerpc: Use generic xfer to guest work function
From: Vishal Chourasia @ 2026-07-09 6:29 UTC (permalink / raw)
To: maddy
Cc: peterz, frederic, npiggin, mpe, chleroy, sshegde, amachhiw,
vaibhav, harshpb, gautam, linuxppc-dev, kvm, linux-kernel,
Vishal Chourasia
This series fixes a KVM scheduling bug on Book3S HV where a guest VM
under a cpu.max bandwidth limit can run arbitrarily past its quota and
then appear frozen for minutes afterwards.
== Problem ==
Since commit 2cd571245b43 ("sched/fair: Add related data structure for
task based throttle"), merged in v6.18, CFS bandwidth throttling no
longer dequeues a task directly. Instead it queues a task_work item via
task_work_add(..., TWA_RESUME), sets TIF_NOTIFY_RESUME, and relies on
that work running on the return path to actually dequeue the task.
The powerpc KVM run loops only test TIF_SIGPENDING and TIF_NEED_RESCHED
before re-entering the guest; TIF_NOTIFY_RESUME is never checked. For a
CPU-bound guest that generates few KVM exits back to userspace, the vCPU
thread never returns to user mode, so the deferred throttle task_work
never runs. The guest keeps running unchecked while its
runtime_remaining goes increasingly negative, and once it finally does
exit to userspace it is legitimately throttled for minutes while the
accrued debt is repaid at the bandwidth-timer replenishment rate.
The generic xfer-to-guest-mode infrastructure (commit 935ace2fb5cc,
"entry: Provide infrastructure for work before transitioning to guest
mode") exists precisely to handle this kind of work before each guest
entry. A full trace-backed root-cause analysis was posted with v1 [2].
== Fix ==
Opt powerpc KVM into VIRT_XFER_TO_GUEST_WORK and use the generic
xfer_to_guest_mode helpers to check for and handle pending guest-mode
work (reschedule, signals, and TIF_NOTIFY_RESUME task_work such as the
deferred CFS throttle) on every guest re-entry:
- Book3S HV: both run loops — kvmhv_run_single_vcpu() for POWER9+ and
kvmppc_run_vcpu() for pre-POWER9.
- Book3S PR and BookE: the common kvmppc_prepare_to_enter(), which
likewise only checked need_resched()/signal_pending().
== Changes from v2 ==
- Add xfer_to_guest_mode_prepare() check for pending rcuog wakeup before
entering guest
- While at it patch all exit points for kvmhv_run_single_vcpu() via
done label matching the corresponding trace_kvmppc_run_vcpu_enter()
with trace_kvmppc_run_vcpu_exit() tracepoint.
- Include posix cpu timer enablement patch by Shrikanth as part of this
series
== Changes from v1 ==
- Extend the fix beyond Book3S HV to the shared powerpc KVM entry path:
also convert the common kvmppc_prepare_to_enter() used by Book3S PR
and BookE. (Shrikanth Shegde)
- Move "select VIRT_XFER_TO_GUEST_WORK" from KVM_BOOK3S_64_HV up to the
common "config KVM" so every powerpc KVM variant gets the
infrastructure.
- Drop the redundant signal_pending() recheck and its sigpend label in
kvmhv_run_single_vcpu(); xfer_to_guest_mode_work_pending() is a
superset of it.
- Preserve the E500 CONFIG_KVM_EXIT_TIMING histogram on the signal path
via an explicit kvmppc_set_exit_type(SIGNAL_EXITS).
[1] https://lore.kernel.org/all/20250421102837.78515-2-sshegde@linux.ibm.com/
[2] https://lore.kernel.org/all/20260626105449.2897924-2-vishalc@linux.ibm.com/
Shrikanth Hegde (1):
powerpc: enable to run posix cpu timers in task context
Vishal Chourasia (1):
KVM: powerpc: Use generic xfer to guest work function
arch/powerpc/Kconfig | 1 +
arch/powerpc/kvm/Kconfig | 1 +
arch/powerpc/kvm/book3s_hv.c | 39 +++++++++++++++++++++++-------------
arch/powerpc/kvm/powerpc.c | 37 +++++++++++++++++++++++++---------
4 files changed, 55 insertions(+), 23 deletions(-)
--
2.54.0
^ permalink raw reply
* Re: [PATCH v16 02/18] syscall_user_dispatch: Introduce a weak fallback for arch_syscall_is_vdso_sigreturn()
From: Jinjie Ruan @ 2026-07-09 6:30 UTC (permalink / raw)
To: Mark Rutland
Cc: oleg, richard.henderson, mattst88, linmag7, linux,
catalin.marinas, will, kees, guoren, chenhuacai, kernel, geert,
tsbogend, James.Bottomley, deller, maddy, mpe, npiggin, chleroy,
pjw, palmer, aou, alex, hca, gor, agordeev, borntraeger, svens,
ysato, dalias, glaubitz, richard, anton.ivanov, johannes, luto,
tglx, mingo, bp, dave.hansen, hpa, chris, jcmvbkbc, peterz, wad,
thuth, ada.coupriediaz, kevin.brodsky, linusw, yeoreum.yun, song,
james.morse, anshuman.khandual, broonie, liqiang01, pengcan,
ryan.roberts, yangtiezhu, sshegde, mchauras, austin.kim, jchrist,
arnd, thomas.weissschuh, sohil.mehta, andrew.cooper3, jgross, kas,
x86, linux-alpha, linux-kernel, linux-arm-kernel, linux-mm,
linux-csky, loongarch, linux-m68k, linux-mips, linux-parisc,
linuxppc-dev, linux-riscv, linux-s390, linux-sh, linux-um
In-Reply-To: <akegeU59jSK9t7FX@J2N7QTR9R3>
On 7/3/2026 7:43 PM, Mark Rutland wrote:
> On Mon, Jun 29, 2026 at 09:06:00PM +0800, Jinjie Ruan wrote:
>> Currently, multiple architectures (LoongArch, RISC-V, S390, Powerpc)
>> provide identical stubs for arch_syscall_is_vdso_sigreturn() that simply
>> return false. This results in redundant boilerplate code across the tree.
>>
>> Introduce a default __weak implementation of
>> arch_syscall_is_vdso_sigreturn() directly in syscall_user_dispatch.c that
>> returns false. This allows architectures that do not utilize a vDSO
>> sigreturn to entirely drop their redundant inline definitions.
>>
>> Architectures requiring a specialized check (such as x86) will continue to
>> override this fallback with their strong symbol definitions.
>>
>> Clean up the redundant implementations in loongarch, riscv, s390
>> and powerpc.
>
>> +bool __weak arch_syscall_is_vdso_sigreturn(struct pt_regs *regs)
>> +{
>> + return false;
>> +}
>
> If we need this, please make it:
>
> #ifndef arch_syscall_is_vdso_sigreturn
> static inline bool arch_syscall_is_vdso_sigreturn(struct pt_regs *regs)
> {
> return false;
> }
> #endif
>
> ... and require that architectures which need this provide a CPP
> definition.
>
> The use of __weak is generally problematic, as it prevents the compiler
> form being able to elide code, and gets in the way of symbol resolution.
> It's perfectly fine to require that architectures need to provide a CPP
> definition alongside their own implementation of this function.
>
> That said, as per my comment on v15, I'd prefer that for now we DO NOT
> enable syscall user dispatch on arm64, and we first make it possible for
> architecture to express whether or not they support that, even if they
> use GENERIC_ENTRY. That might mean this patch isn't necessary right now.
That sounds good. First, focus on switching to generic entry, without
having to implement syscall user dispatch immediately.
>
> [1] https://lore.kernel.org/linux-arm-kernel/akZgV0Y4YAmB43_g@J2N7QTR9R3.cambridge.arm.com/
>
> Mark.
>
^ permalink raw reply
* Re: [PATCH v2] powerpc/64s: Clarify copy_and_flush() cache sync loop comment
From: Nikhil Kumar Singh @ 2026-07-09 6:10 UTC (permalink / raw)
To: Mukesh Kumar Chaurasiya
Cc: linuxppc-dev, linux-kernel, maddy, mpe, npiggin, chleroy, adityag,
mahesh
In-Reply-To: <86bee101-6bd4-45a6-a633-34f2a52d5ace@linux.ibm.com>
Hi Mukesh,
Ideally, checkpatch should take care of this, I see no checkpatch errors.
```
./scripts/checkpatch.pl
v2-0001-powerpc-64s-Clarify-copy_and_flush-cache-sync-loo.patch
total: 0 errors, 0 warnings, 26 lines checked
v2-0001-powerpc-64s-Clarify-copy_and_flush-cache-sync-loo.patch has no
obvious style problems and is ready for submission.
```
Thanks
~ Nikhil
On 07/07/26 12:48, Nikhil Kumar Singh wrote:
> Hi Mukesh,
>
> Thanks for reviewing the patch and for the suggestion.
> I'd prefer to keep the current alignment as I find it a bit more
> readable.
>
> Thanks,
> ~ Nikhil
>
> On 07/07/26 11:08, Mukesh Kumar Chaurasiya wrote:
>> On Mon, Jul 06, 2026 at 01:57:08PM +0530, Nikhil Kumar Singh wrote:
>>> The value loaded into r0 in copy_and_flush() represents the number of
>>> 8-byte words processed between cache synchronization operations.
>>>
>>> The existing comment refers to cache line size, which can make it
>>> appear
>>> that the value is a cache line size in bytes rather than a loop count.
>>> Clarify the comment to explain that the loop processes 8 words (64
>>> bytes)
>>> per cache synchronization iteration, and that increasing the value
>>> would
>>> skip cache maintenance for intermediate cache lines.
>>>
>>> This is a comment-only change with no functional impact.
>>>
>>> Signed-off-by: Nikhil Kumar Singh <nikhilks@linux.ibm.com>
>>> ---
>>> arch/powerpc/kernel/head_64.S | 20 ++++++++++++--------
>>> 1 file changed, 12 insertions(+), 8 deletions(-)
>>>
>>> diff --git a/arch/powerpc/kernel/head_64.S
>>> b/arch/powerpc/kernel/head_64.S
>>> index 63432a33ec49..a54f6f979173 100644
>>> --- a/arch/powerpc/kernel/head_64.S
>>> +++ b/arch/powerpc/kernel/head_64.S
>>> @@ -713,14 +713,18 @@ p_end: .8byte _end - copy_to_here
>>> _GLOBAL(copy_and_flush)
>>> addi r5,r5,-8
>>> addi r6,r6,-8
>>> -4: li r0,8 /* Use the smallest common */
>>> - /* denominator cache line */
>>> - /* size. This results in */
>>> - /* extra cache line flushes */
>>> - /* but operation is correct. */
>>> - /* Can't get cache line size */
>>> - /* from NACA as it is being */
>>> - /* moved too. */
>>> +4: li r0,8 /* r0 is the number of 8-byte words */
>>> + /* to copy per cache sync iteration. */
>>> + /* 8 words * 8 bytes = 64 bytes. 64B is */
>>> + /* the current default cache line size. */
>>> + /* This is a loop count, not a byte */
>>> + /* count. Increasing it may skip */
>>> + /* dcbst/icbi for lines in between and */
>>> + /* leave stale instructions in icache. */
>>> + /* This results in extra cache line */
>>> + /* flushes but operation is correct. */
>>> + /* Can't get cache line size from NACA */
>>> + /* as it is being moved too. */
>> The comment looks fine. But the alignement of the comment is still
>> different from earlier. Keep the alignment same as earlier, it looks a
>> little nicer.
>>
>> Regards,
>> Mukesh
>>> mtctr r0 /* put # words/line in ctr */
>>> 3: addi r6,r6,8 /* copy a cache line */
>>> --
>>> 2.43.5
>>>
^ permalink raw reply
* Re: [PATCH v2] powerpc/powernv: Cache OPAL check_token() results
From: Nikhil Kumar Singh @ 2026-07-09 6:04 UTC (permalink / raw)
To: Shivang Upadhyay, linuxppc-dev; +Cc: Sourabh Jain
In-Reply-To: <fa47cb85e275b9374efbf0f992281fa196ddf8ec.camel@linux.ibm.com>
[-- Attachment #1: Type: text/plain, Size: 940 bytes --]
I discussed this with Sourabh, and we confirmed this function is
actually not on a hot path. However, he is open to the idea. Please take
a look and see if adding unlikely() provides a meaningful benefit here,
or if it falls into premature optimization.
Thanks
~ Nikhil
On 08/07/26 14:02, Shivang Upadhyay wrote:
> On Tue, 2026-07-07 at 12:36 +0530, Nikhil Kumar Singh wrote:
>> You can wrap the if (token_cache[token] == SUPP_UNKNOWN) check, as
>> well as the if (token > OPAL_LAST) check above it, with unlikely(),
>> since this function is in the hot path. Out-of-bounds tokens are very
>> rare, and after the cache is populated, the SUPP_UNKNOWN condition is
>> expected to be false in almost all cases. Wrapping these checks with
>> unlikely() will help the compiler optimize branch
>
> Thanks for your suggestion, but Which hotpath code is calling this?
> we can maybe change that hotpath to not do rtas calls.
>
> ~Shivang.
[-- Attachment #2: Type: text/html, Size: 1957 bytes --]
^ permalink raw reply
* Re: [PATCH V16 4/7] rust/powerpc: Set min rustc version for powerpc
From: Madhavan Srinivasan @ 2026-07-09 6:00 UTC (permalink / raw)
To: Miguel Ojeda, Mukesh Kumar Chaurasiya
Cc: mpe, npiggin, chleroy, peterz, jpoimboe, jbaron, aliceryhl,
rostedt, ardb, ojeda, boqun, gary, bjorn3_gh, lossin, a.hindborg,
tmgross, dakr, nathan, nick.desaulniers+lkml, morbo, justinstitt,
daniel.almeida, acourbot, fujita.tomonori, gregkh, prafulrai522,
tamird, kees, lyude, airlied, linuxppc-dev, linux-kernel,
rust-for-linux, llvm
In-Reply-To: <CANiq72=EVsTFkQK3LWrT2Gs5UKKViemGkv1WuitY6NOEtvcsNg@mail.gmail.com>
On 6/26/26 3:47 PM, Miguel Ojeda wrote:
> On Fri, Jun 26, 2026 at 8:50 AM Mukesh Kumar Chaurasiya
> <mkchauras@gmail.com> wrote:
>> This corresponds to a floating point instruction. Although we have
>> disabled fpu, vsx and altivec we are still generating these. Seems like
>> we have some issue with the rustc here. I didn't dive deeper into this.
>>
>> Is it ok if we keep 1.95 as base for powerpc64le?
> It may be that it is generated but unused (we have that issue
> elsewhere -- see e.g. `rust/compiler_builtins.rs`).
>
> Having said that, if 1.95 is fine for the powerpc arch maintainers,
> then that is fine (s390 also used a higher minimum -- please see the
> recent change in `scripts/min-tool-version.sh`).
Missed to response here. Yes I think 1.95 is fine and have requested
to post a v17 which is already done. Will pull that for the -next
Maddy
^ permalink raw reply
* [PATCH] KVM: PPC: Book3S HV: Set irqfd->producer only on success
From: leixiang @ 2026-07-09 5:57 UTC (permalink / raw)
Cc: seanjc, pbonzini, stable, leixiang, Madhavan Srinivasan,
Nicholas Piggin, Michael Ellerman, Christophe Leroy (CS GROUP),
Suresh Warrier, Paul Mackerras, linuxppc-dev, kvm, linux-kernel
In-Reply-To: <ak59frQUBl9Gs3Qn@google.com>
Set irqfd->producer only after kvmppc_set_passthru_irq() succeeds to
avoid leaving a dangling pointer on failure. The bypass manager does
not register a failed producer, so the pointer is never cleared.
Fixes: c57875f5f9be ("KVM: PPC: Book3S HV: Enable IRQ bypass")
Suggested-by: Sean Christopherson <seanjc@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: leixiang <leixiang@kylinos.cn>
---
arch/powerpc/kvm/book3s_hv.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 61dbeea317f3..ff7b25629125 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -6111,12 +6111,12 @@ static int kvmppc_irq_bypass_add_producer_hv(struct irq_bypass_consumer *cons,
struct kvm_kernel_irqfd *irqfd =
container_of(cons, struct kvm_kernel_irqfd, consumer);
- irqfd->producer = prod;
-
ret = kvmppc_set_passthru_irq(irqfd->kvm, prod->irq, irqfd->gsi);
if (ret)
pr_info("kvmppc_set_passthru_irq (irq %d, gsi %d) fails: %d\n",
prod->irq, irqfd->gsi, ret);
+ else
+ irqfd->producer = prod;
return ret;
}
--
2.45.0
^ permalink raw reply related
* Re: [PATCH] MAINTAINERS: Demote myself to reviewer
From: Madhavan Srinivasan @ 2026-07-09 5:18 UTC (permalink / raw)
To: Michael Ellerman, Ritesh Harjani (IBM), linuxppc-dev,
Christophe Leroy (CS GROUP)
In-Reply-To: <2847c36f-aba7-4458-8767-7617e13058bf@ellerman.id.au>
On 6/24/26 12:48 PM, Michael Ellerman wrote:
> On 11/5/26 11:00 am, Ritesh Harjani (IBM) wrote:
>> Michael Ellerman <mpe@ellerman.id.au> writes:
>>> LINUX FOR POWERPC (32-BIT AND 64-BIT)
>>> M: Madhavan Srinivasan <maddy@linux.ibm.com>
>>> -M: Michael Ellerman <mpe@ellerman.id.au>
>>
>> Thanks for more than a decade of maintainership of Linux on PowerPC tree
>> :)
>
> Thanks Ritesh.
>
> Maddy can you pick this one up please :)
Sure will do
Maddy
>
> cheers
^ permalink raw reply
* Re: [patch 03/18] entry: Provide [syscall_]enter_from_user_mode_randomize_stack()
From: Jinjie Ruan @ 2026-07-09 3:46 UTC (permalink / raw)
To: Thomas Gleixner, LKML
Cc: Peter Zijlstra, Michael Ellerman, Shrikanth Hegde, linuxppc-dev,
Kees Cook, Huacai Chen, loongarch, Paul Walmsley, Palmer Dabbelt,
linux-riscv, Sven Schnelle, linux-s390, x86, Mark Rutland,
Andy Lutomirski, Oleg Nesterov, Richard Henderson, Russell King,
Catalin Marinas, Guo Ren, Geert Uytterhoeven, Thomas Bogendoerfer,
Helge Deller, Yoshinori Sato, Richard Weinberger, Chris Zankel,
linux-arm-kernel, linux-alpha, linux-csky, linux-m68k, linux-mips,
linux-parisc, linux-sh, linux-um, Arnd Bergmann, Vineet Gupta,
Will Deacon, Brian Cain, Michal Simek, Dinh Nguyen,
David S. Miller, Andreas Larsson, linux-snps-arc, linux-hexagon,
linux-openrisc, sparclinux, linux-arch, Michal Suchánek,
Jonathan Corbet, linux-doc
In-Reply-To: <20260707190253.816918647@kernel.org>
On 7/8/2026 3:06 AM, Thomas Gleixner wrote:
> Randomizing the syscall stack can only happen after state is established
> via enter_from_user_mode() or syscall_enter_from_user_mode(). The earlier
> it happens the better.
>
> Provide two new macros to consolidate that:
>
> - enter_from_user_mode_randomize_stack()
> enter_from_user_mode();
> add_random_kstack_offset_irqsoff();
>
> - syscall_enter_from_user_mode_randomize_stack()
> enter_from_user_mode_randomize_stack();
> syscall_enter_from_user_mode_work();
>
> to reduce boiler plate code.
>
> Those are macros and not inline functions as the latter would limit the
> stack randomization scope to the inline function itself.
>
> Signed-off-by: Thomas Gleixner <tglx@kernel.org>
> ---
> include/linux/entry-common.h | 56 +++++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 56 insertions(+)
>
> --- a/include/linux/entry-common.h
> +++ b/include/linux/entry-common.h
> @@ -6,6 +6,7 @@
> #include <linux/irq-entry-common.h>
> #include <linux/livepatch.h>
> #include <linux/ptrace.h>
> +#include <linux/randomize_kstack.h>
> #include <linux/resume_user_mode.h>
> #include <linux/seccomp.h>
> #include <linux/sched.h>
> @@ -150,6 +151,61 @@ static __always_inline long syscall_ente
> }
>
> /**
> + * enter_from_user_mode_randomize_stack - Establish state and add stack randomization
> + * before invoking syscall_enter_from_user_mode_work()
> + * @regs: Pointer to currents pt_regs
> + *
> + * Invoked from architecture specific syscall entry code with interrupts
> + * disabled. The calling code has to be non-instrumentable. When the function
> + * returns all state is correct, interrupts are still disabled and the
> + * subsequent functions can be instrumented.
> + *
> + * Implemented as a macro so that the stack randomization is effective
> + * throughout the function in which it is invoked. An inline would only make it
> + * effective in the scope of the inline function.
> + */
> +#define enter_from_user_mode_randomize_stack(regs) \
> +do { \
> + enter_from_user_mode(regs); \
> + instrumentation_begin(); \
> + add_random_kstack_offset_irqsoff(); \
> + instrumentation_end(); \
> +} while (0)
Perhaps this new function can also be reused when the ARM64 is switched
to the generic entry as the irq also disabled now.
--- a/arch/arm64/kernel/entry-common.c
+++ b/arch/arm64/kernel/entry-common.c
@@ -64,7 +64,7 @@ static void noinstr arm64_exit_to_kernel_mode(struct
pt_regs *regs,
static __always_inline void arm64_syscall_enter_from_user_mode(struct
pt_regs *regs)
{
- enter_from_user_mode(regs);
+ enter_from_user_mode_randomize_stack(regs);
mte_disable_tco_entry(current);
sme_enter_from_user_mode();
}
diff --git a/arch/arm64/kernel/syscall.c b/arch/arm64/kernel/syscall.c
index e0a98fac3b85..42ac02573b66 100644
--- a/arch/arm64/kernel/syscall.c
+++ b/arch/arm64/kernel/syscall.c
@@ -6,7 +6,6 @@
#include <linux/errno.h>
#include <linux/nospec.h>
#include <linux/ptrace.h>
-#include <linux/randomize_kstack.h>
#include <linux/syscalls.h>
#include <asm/debug-monitors.h>
@@ -42,8 +41,6 @@ static void invoke_syscall(struct pt_regs *regs,
unsigned int scno,
{
long ret;
- add_random_kstack_offset();
-
if (likely(scno < sc_nr)) {
syscall_fn_t syscall_fn;
syscall_fn = syscall_table[array_index_nospec(scno, sc_nr)];
> +
> +/**
> + * syscall_enter_from_user_mode_randomize_stack - Establish state and check and handle work
> + * before invoking a syscall
> + * @regs: Pointer to currents pt_regs
> + * @syscall: The syscall number
> + *
> + * Invoked from architecture specific syscall entry code with interrupts
> + * disabled. The calling code has to be non-instrumentable. When the
> + * function returns all state is correct, interrupts are enabled and the
> + * subsequent functions can be instrumented.
> + *
> + * This is the combination of enter_from_user_mode_randomize_stack() and
> + * syscall_enter_from_user_mode_work() to be used when there is no
> + * architecture specific work to be done between the two.
> + *
> + * Returns: The original or a modified syscall number. See
> + * syscall_enter_from_user_mode_work() for further explanation.
> + *
> + * Implemented as a macro to make stack randomization effective in the calling
> + * scope.
> + */
> +#define syscall_enter_from_user_mode_randomize_stack(regs, syscall) \
> +({ \
> + enter_from_user_mode_randomize_stack(regs); \
> + \
> + instrumentation_begin(); \
> + local_irq_enable(); \
> + long _ret = syscall_enter_from_user_mode_work(regs, syscall); \
> + instrumentation_end(); \
> + \
> + _ret; \
> +})
> +
> +/**
> * syscall_enter_from_user_mode - Establish state and check and handle work
> * before invoking a syscall
> * @regs: Pointer to currents pt_regs
>
^ permalink raw reply related
* Re: [patch 10/18] entry: Use syscall number instead of rereading it
From: Jinjie Ruan @ 2026-07-09 2:55 UTC (permalink / raw)
To: Thomas Gleixner, LKML
Cc: Peter Zijlstra, Michael Ellerman, Shrikanth Hegde, linuxppc-dev,
Kees Cook, Huacai Chen, loongarch, Paul Walmsley, Palmer Dabbelt,
linux-riscv, Sven Schnelle, linux-s390, x86, Mark Rutland,
Andy Lutomirski, Oleg Nesterov, Richard Henderson, Russell King,
Catalin Marinas, Guo Ren, Geert Uytterhoeven, Thomas Bogendoerfer,
Helge Deller, Yoshinori Sato, Richard Weinberger, Chris Zankel,
linux-arm-kernel, linux-alpha, linux-csky, linux-m68k, linux-mips,
linux-parisc, linux-sh, linux-um, Arnd Bergmann, Vineet Gupta,
Will Deacon, Brian Cain, Michal Simek, Dinh Nguyen,
David S. Miller, Andreas Larsson, linux-snps-arc, linux-hexagon,
linux-openrisc, sparclinux, linux-arch, Michal Suchánek,
Jonathan Corbet, linux-doc
In-Reply-To: <20260707190254.181086755@kernel.org>
On 7/8/2026 3:06 AM, Thomas Gleixner wrote:
> rseq_syscall_enter_work() is invoked before the syscall number can be
> modified. So there is no point in rereading it from pt_regs.
>
> Signed-off-by: Thomas Gleixner <tglx@kernel.org>
> ---
> include/linux/entry-common.h | 9 +++++----
> 1 file changed, 5 insertions(+), 4 deletions(-)
>
> --- a/include/linux/entry-common.h
> +++ b/include/linux/entry-common.h
> @@ -70,9 +70,10 @@ static inline void syscall_enter_audit(s
> }
> }
>
> -static __always_inline long syscall_trace_enter(struct pt_regs *regs, unsigned long work)
> +static __always_inline long syscall_trace_enter(struct pt_regs *regs, unsigned long work,
> + long syscall)
> {
> - long syscall, ret = 0;
> + long ret = 0;
>
> /*
> * Handle Syscall User Dispatch. This must comes first, since
> @@ -90,7 +91,7 @@ static __always_inline long syscall_trac
> * through hrtimer_interrupt().
> */
> if (work & SYSCALL_WORK_SYSCALL_RSEQ_SLICE)
> - rseq_syscall_enter_work(syscall_get_nr(current, regs));
> + rseq_syscall_enter_work(syscall);
>
> /* Handle ptrace */
> if (work & (SYSCALL_WORK_SYSCALL_TRACE | SYSCALL_WORK_SYSCALL_EMU)) {
> @@ -145,7 +146,7 @@ static __always_inline long syscall_ente
> unsigned long work = READ_ONCE(current_thread_info()->syscall_work);
>
> if (work & SYSCALL_WORK_ENTER)
> - syscall = syscall_trace_enter(regs, work);
> + syscall = syscall_trace_enter(regs, work, syscall);
Reviewed-by: Jinjie Ruan <ruanjinjie@huawei.com>
>
> return syscall;
> }
>
^ permalink raw reply
* Re: [patch 09/18] entry: Remove syscall_enter_from_user_mode()
From: Jinjie Ruan @ 2026-07-09 2:49 UTC (permalink / raw)
To: Thomas Gleixner, LKML
Cc: Peter Zijlstra, Michael Ellerman, Shrikanth Hegde, linuxppc-dev,
Kees Cook, Huacai Chen, loongarch, Paul Walmsley, Palmer Dabbelt,
linux-riscv, Sven Schnelle, linux-s390, x86, Mark Rutland,
Andy Lutomirski, Oleg Nesterov, Richard Henderson, Russell King,
Catalin Marinas, Guo Ren, Geert Uytterhoeven, Thomas Bogendoerfer,
Helge Deller, Yoshinori Sato, Richard Weinberger, Chris Zankel,
linux-arm-kernel, linux-alpha, linux-csky, linux-m68k, linux-mips,
linux-parisc, linux-sh, linux-um, Arnd Bergmann, Vineet Gupta,
Will Deacon, Brian Cain, Michal Simek, Dinh Nguyen,
David S. Miller, Andreas Larsson, linux-snps-arc, linux-hexagon,
linux-openrisc, sparclinux, linux-arch, Michal Suchánek,
Jonathan Corbet, linux-doc
In-Reply-To: <20260707190254.132654198@kernel.org>
On 7/8/2026 3:06 AM, Thomas Gleixner wrote:
> All architecture use either:
>
> nr = enter_from_user_mode_randomize_stack(regs, nr);
>
> or
>
> enter_from_user_mode_randomize_stack(regs);
> nr = syscall_enter_from_user_mode_work(regs, nr);
>
> Remove the now unused function.
>
> Signed-off-by: Thomas Gleixner <tglx@kernel.org>
> ---
> Documentation/core-api/entry.rst | 17 +++++++++-------
> include/linux/entry-common.h | 40 +++------------------------------------
> include/linux/irq-entry-common.h | 6 ++---
> 3 files changed, 17 insertions(+), 46 deletions(-)
>
> --- a/Documentation/core-api/entry.rst
> +++ b/Documentation/core-api/entry.rst
> @@ -68,7 +68,7 @@ low-level C code must not be instrumente
> noinstr void syscall(struct pt_regs *regs, int nr)
> {
> arch_syscall_enter(regs);
> - nr = syscall_enter_from_user_mode(regs, nr);
> + nr = syscall_enter_from_user_mode_randomize_stack(regs, nr);
>
> instrumentation_begin();
> if (!invoke_syscall(regs, nr) && nr != -1)
> @@ -78,12 +78,14 @@ low-level C code must not be instrumente
> syscall_exit_to_user_mode(regs);
> }
>
> -syscall_enter_from_user_mode() first invokes enter_from_user_mode() which
> -establishes state in the following order:
> +syscall_enter_from_user_mode_randomize_stack() first invokes
> +enter_from_user_mode_randomize_stack() which establishes state in the
> +following order:
>
> * Lockdep
> * RCU / Context tracking
> * Tracing
> + * Apply stack randomization
>
> and then invokes the various entry work functions like ptrace, seccomp, audit,
> syscall tracing, etc. After all that is done, the instrumentable invoke_syscall
> @@ -99,10 +101,11 @@ that it invokes exit_to_user_mode() whic
> * RCU / Context tracking
> * Lockdep
>
> -syscall_enter_from_user_mode() and syscall_exit_to_user_mode() are also
> -available as fine grained subfunctions in cases where the architecture code
> -has to do extra work between the various steps. In such cases it has to
> -ensure that enter_from_user_mode() is called first on entry and
> +syscall_enter_from_user_mode_randomize_stack() and
> +syscall_exit_to_user_mode() are also available as fine grained subfunctions
> +in cases where the architecture code has to do extra work between the
> +various steps. In such cases it has to ensure that
> +enter_from_user_mode_randomize_stack() is called first on entry and
> exit_to_user_mode() is called last on exit.
>
> Do not nest syscalls. Nested syscalls will cause RCU and/or context tracking
> --- a/include/linux/entry-common.h
> +++ b/include/linux/entry-common.h
> @@ -19,7 +19,7 @@
> #endif
>
> /*
> - * SYSCALL_WORK flags handled in syscall_enter_from_user_mode()
> + * SYSCALL_WORK flags handled in syscall_enter_from_user_mode_work()
> */
> #define SYSCALL_WORK_ENTER (SYSCALL_WORK_SECCOMP | \
> SYSCALL_WORK_SYSCALL_TRACEPOINT | \
> @@ -205,42 +205,10 @@ do { \
> _ret; \
> })
>
> -/**
> - * syscall_enter_from_user_mode - Establish state and check and handle work
> - * before invoking a syscall
> - * @regs: Pointer to currents pt_regs
> - * @syscall: The syscall number
> - *
> - * Invoked from architecture specific syscall entry code with interrupts
> - * disabled. The calling code has to be non-instrumentable. When the
> - * function returns all state is correct, interrupts are enabled and the
> - * subsequent functions can be instrumented.
> - *
> - * This is the combination of enter_from_user_mode() and
> - * syscall_enter_from_user_mode_work() to be used when there is no
> - * architecture specific work to be done between the two.
> - *
> - * Returns: The original or a modified syscall number. See
> - * syscall_enter_from_user_mode_work() for further explanation.
> - */
> -static __always_inline long syscall_enter_from_user_mode(struct pt_regs *regs, long syscall)
> -{
> - long ret;
> -
> - enter_from_user_mode(regs);
> -
> - instrumentation_begin();
> - local_irq_enable();
> - ret = syscall_enter_from_user_mode_work(regs, syscall);
> - instrumentation_end();
> -
> - return ret;
> -}
> -
> /*
> - * If SYSCALL_EMU is set, then the only reason to report is when
> - * SINGLESTEP is set (i.e. PTRACE_SYSEMU_SINGLESTEP). This syscall
> - * instruction has been already reported in syscall_enter_from_user_mode().
> + * If SYSCALL_EMU is set, then the only reason to report is when SINGLESTEP is
> + * set (i.e. PTRACE_SYSEMU_SINGLESTEP). This syscall instruction has been
> + * already reported in syscall_enter_from_user_mode_work().
> */
> static __always_inline bool report_single_step(unsigned long work)
> {
> --- a/include/linux/irq-entry-common.h
> +++ b/include/linux/irq-entry-common.h
> @@ -49,9 +49,9 @@
> * Defaults to an empty implementation. Can be replaced by architecture
> * specific code.
> *
> - * Invoked from syscall_enter_from_user_mode() in the non-instrumentable
> - * section. Use __always_inline so the compiler cannot push it out of line
> - * and make it instrumentable.
> + * Invoked from enter_from_user_mode_syscall_and_randomize_stack() in the
> + * non-instrumentable section. Use __always_inline so the compiler cannot push
> + * it out of line and make it instrumentable.
Reviewed-by: Jinjie Ruan <ruanjinjie@huawei.com>
> */
> static __always_inline void arch_enter_from_user_mode(struct pt_regs *regs);
>
>
^ permalink raw reply
* Re: [patch 07/18] s390/syscall: Use enter_from_user_mode_randomize_stack()
From: Jinjie Ruan @ 2026-07-09 2:46 UTC (permalink / raw)
To: Thomas Gleixner, LKML
Cc: Peter Zijlstra, Sven Schnelle, linux-s390, Michael Ellerman,
Shrikanth Hegde, linuxppc-dev, Kees Cook, Huacai Chen, loongarch,
Paul Walmsley, Palmer Dabbelt, linux-riscv, x86, Mark Rutland,
Andy Lutomirski, Oleg Nesterov, Richard Henderson, Russell King,
Catalin Marinas, Guo Ren, Geert Uytterhoeven, Thomas Bogendoerfer,
Helge Deller, Yoshinori Sato, Richard Weinberger, Chris Zankel,
linux-arm-kernel, linux-alpha, linux-csky, linux-m68k, linux-mips,
linux-parisc, linux-sh, linux-um, Arnd Bergmann, Vineet Gupta,
Will Deacon, Brian Cain, Michal Simek, Dinh Nguyen,
David S. Miller, Andreas Larsson, linux-snps-arc, linux-hexagon,
linux-openrisc, sparclinux, linux-arch, Michal Suchánek,
Jonathan Corbet, linux-doc
In-Reply-To: <20260707190254.030598804@kernel.org>
On 7/8/2026 3:06 AM, Thomas Gleixner wrote:
> enter_from_user_mode_randomize_stack() replaces enter_from_user_mode() and
> the subsequent invocation of add_random_kstack_offset_irqsoff().
>
> As a bonus this avoids the overhead of get/put_cpu_var() in
> add_random_kstack_offset().
>
> No functional change.
>
> Signed-off-by: Thomas Gleixner <tglx@kernel.org>
> Cc: Sven Schnelle <svens@linux.ibm.com>
> Cc: linux-s390@vger.kernel.org
> ---
> arch/s390/kernel/syscall.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> --- a/arch/s390/kernel/syscall.c
> +++ b/arch/s390/kernel/syscall.c
> @@ -97,8 +97,8 @@ void noinstr __do_syscall(struct pt_regs
> {
> unsigned long nr;
>
> - enter_from_user_mode(regs);
> - add_random_kstack_offset();
> + enter_from_user_mode_randomize_stack(regs);
The #include <linux/randomize_kstack.h> can be removed.
> +
> regs->psw = get_lowcore()->svc_old_psw;
> regs->int_code = get_lowcore()->svc_int_code;
> update_timer_sys();
>
^ permalink raw reply
* Re: [patch 08/18] x86/syscall: Use [syscall_]enter_from_user_mode_randomize_stack()
From: Jinjie Ruan @ 2026-07-09 2:44 UTC (permalink / raw)
To: Thomas Gleixner, LKML
Cc: Peter Zijlstra, x86, Michael Ellerman, Shrikanth Hegde,
linuxppc-dev, Kees Cook, Huacai Chen, loongarch, Paul Walmsley,
Palmer Dabbelt, linux-riscv, Sven Schnelle, linux-s390,
Mark Rutland, Andy Lutomirski, Oleg Nesterov, Richard Henderson,
Russell King, Catalin Marinas, Guo Ren, Geert Uytterhoeven,
Thomas Bogendoerfer, Helge Deller, Yoshinori Sato,
Richard Weinberger, Chris Zankel, linux-arm-kernel, linux-alpha,
linux-csky, linux-m68k, linux-mips, linux-parisc, linux-sh,
linux-um, Arnd Bergmann, Vineet Gupta, Will Deacon, Brian Cain,
Michal Simek, Dinh Nguyen, David S. Miller, Andreas Larsson,
linux-snps-arc, linux-hexagon, linux-openrisc, sparclinux,
linux-arch, Michal Suchánek, Jonathan Corbet, linux-doc
In-Reply-To: <20260707190254.079478122@kernel.org>
On 7/8/2026 3:06 AM, Thomas Gleixner wrote:
> These functions integrate the stack randomization.
>
> syscall_enter_from_user_mode_randomize_stack() has the advantage that the
> randomization happens early right after enter_from_user_mode().
>
> In both cases also the overhead of get/put_cpu_var() in
> add_random_kstack_offset() is avoided.
>
> No functional change.
>
> Signed-off-by: Thomas Gleixner <tglx@kernel.org>
> Cc: x86@kernel.org
> ---
> arch/x86/entry/syscall_32.c | 19 +++++--------------
> arch/x86/entry/syscall_64.c | 3 +--
> arch/x86/include/asm/entry-common.h | 1 -
> 3 files changed, 6 insertions(+), 17 deletions(-)
Reviewed-by: Jinjie Ruan <ruanjinjie@huawei.com>
>
> --- a/arch/x86/entry/syscall_32.c
> +++ b/arch/x86/entry/syscall_32.c
> @@ -142,10 +142,9 @@ static __always_inline bool int80_is_ext
> * int80_is_external() below which calls into the APIC driver.
> * Identical for soft and external interrupts.
> */
> - enter_from_user_mode(regs);
> + enter_from_user_mode_randomize_stack(regs);
>
> instrumentation_begin();
> - add_random_kstack_offset();
>
> /* Validate that this is a soft interrupt to the extent possible */
> if (unlikely(int80_is_external()))
> @@ -210,11 +209,9 @@ DEFINE_FREDENTRY_RAW(int80_emulation)
> {
> int nr;
>
> - enter_from_user_mode(regs);
> + enter_from_user_mode_randomize_stack(regs);
>
> instrumentation_begin();
> - add_random_kstack_offset();
> -
> /*
> * FRED pushed 0 into regs::orig_ax and regs::ax contains the
> * syscall number.
> @@ -252,10 +249,10 @@ DEFINE_FREDENTRY_RAW(int80_emulation)
> * orig_ax, the int return value truncates it. This matches
> * the semantics of syscall_get_nr().
> */
> - nr = syscall_enter_from_user_mode(regs, nr);
> + nr = syscall_enter_from_user_mode_randomize_stack(regs, nr);
> +
> instrumentation_begin();
>
> - add_random_kstack_offset();
> do_syscall_32_irqs_on(regs, nr);
>
> instrumentation_end();
> @@ -268,15 +265,9 @@ static noinstr bool __do_fast_syscall_32
> int nr = syscall_32_enter(regs);
> int res;
>
> - /*
> - * This cannot use syscall_enter_from_user_mode() as it has to
> - * fetch EBP before invoking any of the syscall entry work
> - * functions.
> - */
> - enter_from_user_mode(regs);
> + enter_from_user_mode_randomize_stack(regs);
>
> instrumentation_begin();
> - add_random_kstack_offset();
> local_irq_enable();
> /* Fetch EBP from where the vDSO stashed it. */
> if (IS_ENABLED(CONFIG_X86_64)) {
> --- a/arch/x86/entry/syscall_64.c
> +++ b/arch/x86/entry/syscall_64.c
> @@ -86,10 +86,9 @@ static __always_inline bool do_syscall_x
> /* Returns true to return using SYSRET, or false to use IRET */
> __visible noinstr bool do_syscall_64(struct pt_regs *regs, int nr)
> {
> - nr = syscall_enter_from_user_mode(regs, nr);
> + nr = syscall_enter_from_user_mode_randomize_stack(regs, nr);
>
> instrumentation_begin();
> - add_random_kstack_offset();
>
> if (!do_syscall_x64(regs, nr) && !do_syscall_x32(regs, nr) && nr != -1) {
> /* Invalid system call, but still a system call. */
> --- a/arch/x86/include/asm/entry-common.h
> +++ b/arch/x86/include/asm/entry-common.h
> @@ -2,7 +2,6 @@
> #ifndef _ASM_X86_ENTRY_COMMON_H
> #define _ASM_X86_ENTRY_COMMON_H
>
> -#include <linux/randomize_kstack.h>
> #include <linux/user-return-notifier.h>
>
> #include <asm/nospec-branch.h>
>
^ permalink raw reply
* Re: [patch 07/18] s390/syscall: Use enter_from_user_mode_randomize_stack()
From: Jinjie Ruan @ 2026-07-09 2:39 UTC (permalink / raw)
To: Thomas Gleixner, LKML
Cc: Peter Zijlstra, Sven Schnelle, linux-s390, Michael Ellerman,
Shrikanth Hegde, linuxppc-dev, Kees Cook, Huacai Chen, loongarch,
Paul Walmsley, Palmer Dabbelt, linux-riscv, x86, Mark Rutland,
Andy Lutomirski, Oleg Nesterov, Richard Henderson, Russell King,
Catalin Marinas, Guo Ren, Geert Uytterhoeven, Thomas Bogendoerfer,
Helge Deller, Yoshinori Sato, Richard Weinberger, Chris Zankel,
linux-arm-kernel, linux-alpha, linux-csky, linux-m68k, linux-mips,
linux-parisc, linux-sh, linux-um, Arnd Bergmann, Vineet Gupta,
Will Deacon, Brian Cain, Michal Simek, Dinh Nguyen,
David S. Miller, Andreas Larsson, linux-snps-arc, linux-hexagon,
linux-openrisc, sparclinux, linux-arch, Michal Suchánek,
Jonathan Corbet, linux-doc
In-Reply-To: <20260707190254.030598804@kernel.org>
On 7/8/2026 3:06 AM, Thomas Gleixner wrote:
> enter_from_user_mode_randomize_stack() replaces enter_from_user_mode() and
> the subsequent invocation of add_random_kstack_offset_irqsoff().
>
> As a bonus this avoids the overhead of get/put_cpu_var() in
> add_random_kstack_offset().
>
> No functional change.
>
> Signed-off-by: Thomas Gleixner <tglx@kernel.org>
> Cc: Sven Schnelle <svens@linux.ibm.com>
> Cc: linux-s390@vger.kernel.org
> ---
> arch/s390/kernel/syscall.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> --- a/arch/s390/kernel/syscall.c
> +++ b/arch/s390/kernel/syscall.c
> @@ -97,8 +97,8 @@ void noinstr __do_syscall(struct pt_regs
> {
> unsigned long nr;
>
> - enter_from_user_mode(regs);
> - add_random_kstack_offset();
> + enter_from_user_mode_randomize_stack(regs);
Reviewed-by: Jinjie Ruan <ruanjinjie@huawei.com>
> +
> regs->psw = get_lowcore()->svc_old_psw;
> regs->int_code = get_lowcore()->svc_int_code;
> update_timer_sys();
>
^ permalink raw reply
* Re: [patch 06/18] riscv/syscall: Use syscall_enter_from_user_mode_randomize_stack()
From: Jinjie Ruan @ 2026-07-09 2:38 UTC (permalink / raw)
To: Thomas Gleixner, LKML
Cc: Peter Zijlstra, Paul Walmsley, Palmer Dabbelt, linux-riscv,
Michael Ellerman, Shrikanth Hegde, linuxppc-dev, Kees Cook,
Huacai Chen, loongarch, Sven Schnelle, linux-s390, x86,
Mark Rutland, Andy Lutomirski, Oleg Nesterov, Richard Henderson,
Russell King, Catalin Marinas, Guo Ren, Geert Uytterhoeven,
Thomas Bogendoerfer, Helge Deller, Yoshinori Sato,
Richard Weinberger, Chris Zankel, linux-arm-kernel, linux-alpha,
linux-csky, linux-m68k, linux-mips, linux-parisc, linux-sh,
linux-um, Arnd Bergmann, Vineet Gupta, Will Deacon, Brian Cain,
Michal Simek, Dinh Nguyen, David S. Miller, Andreas Larsson,
linux-snps-arc, linux-hexagon, linux-openrisc, sparclinux,
linux-arch, Michal Suchánek, Jonathan Corbet, linux-doc
In-Reply-To: <20260707190253.974626922@kernel.org>
On 7/8/2026 3:06 AM, Thomas Gleixner wrote:
> syscall_enter_from_user_mode_randomize_stack() replaces
> syscall_enter_from_user_mode() and the subsequent invocation of
> add_random_kstack_offset().
>
> The advantage is that it applies the stack randomization right after
> enter_from_user_mode() and thereby avoids the overhead of get/put_cpu_var()
> as that code is invoked with interrupts disabled.
>
> No functional change.
>
> Signed-off-by: Thomas Gleixner <tglx@kernel.org>
> Cc: Paul Walmsley <pjw@kernel.org>
> Cc: Palmer Dabbelt <palmer@dabbelt.com>
> Cc: linux-riscv@lists.infradead.org
> ---
> arch/riscv/kernel/traps.c | 5 +----
> 1 file changed, 1 insertion(+), 4 deletions(-)
>
> --- a/arch/riscv/kernel/traps.c
> +++ b/arch/riscv/kernel/traps.c
> @@ -7,7 +7,6 @@
> #include <linux/kernel.h>
> #include <linux/init.h>
> #include <linux/irqflags.h>
> -#include <linux/randomize_kstack.h>
> #include <linux/sched.h>
> #include <linux/sched/debug.h>
> #include <linux/sched/signal.h>
> @@ -333,9 +332,7 @@ void do_trap_ecall_u(struct pt_regs *reg
>
> riscv_v_vstate_discard(regs);
>
> - syscall = syscall_enter_from_user_mode(regs, syscall);
> -
> - add_random_kstack_offset();
> + syscall = syscall_enter_from_user_mode_randomize_stack(regs, syscall);
Reviewed-by: Jinjie Ruan <ruanjinjie@huawei.com>
>
> if (syscall >= 0 && syscall < NR_syscalls) {
> syscall = array_index_nospec(syscall, NR_syscalls);
>
^ permalink raw reply
* Re: [patch 05/18] powerpc/syscall: Use syscall_enter_from_user_mode_randomize_stack()
From: Jinjie Ruan @ 2026-07-09 2:38 UTC (permalink / raw)
To: Thomas Gleixner, LKML
Cc: Peter Zijlstra, Michael Ellerman, Shrikanth Hegde, linuxppc-dev,
Kees Cook, Huacai Chen, loongarch, Paul Walmsley, Palmer Dabbelt,
linux-riscv, Sven Schnelle, linux-s390, x86, Mark Rutland,
Andy Lutomirski, Oleg Nesterov, Richard Henderson, Russell King,
Catalin Marinas, Guo Ren, Geert Uytterhoeven, Thomas Bogendoerfer,
Helge Deller, Yoshinori Sato, Richard Weinberger, Chris Zankel,
linux-arm-kernel, linux-alpha, linux-csky, linux-m68k, linux-mips,
linux-parisc, linux-sh, linux-um, Arnd Bergmann, Vineet Gupta,
Will Deacon, Brian Cain, Michal Simek, Dinh Nguyen,
David S. Miller, Andreas Larsson, linux-snps-arc, linux-hexagon,
linux-openrisc, sparclinux, linux-arch, Michal Suchánek,
Jonathan Corbet, linux-doc
In-Reply-To: <20260707190253.918861529@kernel.org>
On 7/8/2026 3:06 AM, Thomas Gleixner wrote:
> syscall_enter_from_user_mode_randomize_stack() replaces
> syscall_enter_from_user_mode() and the subsequent invocation of
> add_random_kstack_offset().
>
> The advantage is that it applies the stack randomization right after
> enter_from_user_mode() and thereby avoids the overhead of get/put_cpu_var()
> as that code is invoked with interrupts disabled.
>
> No functional change.
>
> Signed-off-by: Thomas Gleixner <tglx@kernel.org>
> Cc: Michael Ellerman <mpe@ellerman.id.au>
> Cc: Shrikanth Hegde <sshegde@linux.ibm.com>
> Cc: linuxppc-dev@lists.ozlabs.org
> ---
> arch/powerpc/kernel/syscall.c | 4 +---
> 1 file changed, 1 insertion(+), 3 deletions(-)
>
> --- a/arch/powerpc/kernel/syscall.c
> +++ b/arch/powerpc/kernel/syscall.c
> @@ -2,7 +2,6 @@
>
> #include <linux/compat.h>
> #include <linux/context_tracking.h>
> -#include <linux/randomize_kstack.h>
> #include <linux/entry-common.h>
>
> #include <asm/interrupt.h>
> @@ -19,8 +18,7 @@ notrace long system_call_exception(struc
> long ret;
> syscall_fn f;
>
> - r0 = syscall_enter_from_user_mode(regs, r0);
> - add_random_kstack_offset();
> + r0 = syscall_enter_from_user_mode_randomize_stack(regs, r0);
Reviewed-by: Jinjie Ruan <ruanjinjie@huawei.com>
>
> if (unlikely(r0 >= NR_syscalls)) {
> if (unlikely(trap_is_unsupported_scv(regs))) {
>
^ permalink raw reply
* Re: [patch 04/18] loongarch/syscall: Use syscall_enter_from_user_mode_randomize_stack()
From: Jinjie Ruan @ 2026-07-09 2:37 UTC (permalink / raw)
To: Thomas Gleixner, LKML
Cc: Peter Zijlstra, Huacai Chen, loongarch, Michael Ellerman,
Shrikanth Hegde, linuxppc-dev, Kees Cook, Paul Walmsley,
Palmer Dabbelt, linux-riscv, Sven Schnelle, linux-s390, x86,
Mark Rutland, Andy Lutomirski, Oleg Nesterov, Richard Henderson,
Russell King, Catalin Marinas, Guo Ren, Geert Uytterhoeven,
Thomas Bogendoerfer, Helge Deller, Yoshinori Sato,
Richard Weinberger, Chris Zankel, linux-arm-kernel, linux-alpha,
linux-csky, linux-m68k, linux-mips, linux-parisc, linux-sh,
linux-um, Arnd Bergmann, Vineet Gupta, Will Deacon, Brian Cain,
Michal Simek, Dinh Nguyen, David S. Miller, Andreas Larsson,
linux-snps-arc, linux-hexagon, linux-openrisc, sparclinux,
linux-arch, Michal Suchánek, Jonathan Corbet, linux-doc
In-Reply-To: <20260707190253.865955911@kernel.org>
On 7/8/2026 3:06 AM, Thomas Gleixner wrote:
> syscall_enter_from_user_mode_randomize_stack() replaces
> syscall_enter_from_user_mode() and the subsequent invocation of
> add_random_kstack_offset().
>
> The advantage is that it applies the stack randomization right after
> enter_from_user_mode() and thereby avoids the overhead of get/put_cpu_var()
> as that code is invoked with interrupts disabled.
>
> No functional change.
>
> Signed-off-by: Thomas Gleixner <tglx@kernel.org>
> Cc: Huacai Chen <chenhuacai@kernel.org>
> Cc: loongarch@lists.linux.dev
> ---
> arch/loongarch/kernel/syscall.c | 5 +----
> 1 file changed, 1 insertion(+), 4 deletions(-)
>
> --- a/arch/loongarch/kernel/syscall.c
> +++ b/arch/loongarch/kernel/syscall.c
> @@ -11,7 +11,6 @@
> #include <linux/linkage.h>
> #include <linux/nospec.h>
> #include <linux/objtool.h>
> -#include <linux/randomize_kstack.h>
> #include <linux/syscalls.h>
> #include <linux/unistd.h>
>
> @@ -70,9 +69,7 @@ void noinstr __no_stack_protector do_sys
> regs->orig_a0 = regs->regs[4];
> regs->regs[4] = -ENOSYS;
>
> - nr = syscall_enter_from_user_mode(regs, nr);
> -
> - add_random_kstack_offset();
> + nr = syscall_enter_from_user_mode_randomize_stack(regs, nr);
Reviewed-by: Jinjie Ruan <ruanjinjie@huawei.com>
>
> if (nr < NR_syscalls) {
> syscall_fn = sys_call_table[array_index_nospec(nr, NR_syscalls)];
>
^ permalink raw reply
* Re: [patch 03/18] entry: Provide [syscall_]enter_from_user_mode_randomize_stack()
From: Jinjie Ruan @ 2026-07-09 2:34 UTC (permalink / raw)
To: Thomas Gleixner, LKML
Cc: Peter Zijlstra, Michael Ellerman, Shrikanth Hegde, linuxppc-dev,
Kees Cook, Huacai Chen, loongarch, Paul Walmsley, Palmer Dabbelt,
linux-riscv, Sven Schnelle, linux-s390, x86, Mark Rutland,
Andy Lutomirski, Oleg Nesterov, Richard Henderson, Russell King,
Catalin Marinas, Guo Ren, Geert Uytterhoeven, Thomas Bogendoerfer,
Helge Deller, Yoshinori Sato, Richard Weinberger, Chris Zankel,
linux-arm-kernel, linux-alpha, linux-csky, linux-m68k, linux-mips,
linux-parisc, linux-sh, linux-um, Arnd Bergmann, Vineet Gupta,
Will Deacon, Brian Cain, Michal Simek, Dinh Nguyen,
David S. Miller, Andreas Larsson, linux-snps-arc, linux-hexagon,
linux-openrisc, sparclinux, linux-arch, Michal Suchánek,
Jonathan Corbet, linux-doc
In-Reply-To: <20260707190253.816918647@kernel.org>
On 7/8/2026 3:06 AM, Thomas Gleixner wrote:
> Randomizing the syscall stack can only happen after state is established
> via enter_from_user_mode() or syscall_enter_from_user_mode(). The earlier
> it happens the better.
>
> Provide two new macros to consolidate that:
>
> - enter_from_user_mode_randomize_stack()
> enter_from_user_mode();
> add_random_kstack_offset_irqsoff();
>
> - syscall_enter_from_user_mode_randomize_stack()
> enter_from_user_mode_randomize_stack();
> syscall_enter_from_user_mode_work();
>
> to reduce boiler plate code.
>
> Those are macros and not inline functions as the latter would limit the
> stack randomization scope to the inline function itself.
>
> Signed-off-by: Thomas Gleixner <tglx@kernel.org>
> ---
> include/linux/entry-common.h | 56 +++++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 56 insertions(+)
>
> --- a/include/linux/entry-common.h
> +++ b/include/linux/entry-common.h
> @@ -6,6 +6,7 @@
> #include <linux/irq-entry-common.h>
> #include <linux/livepatch.h>
> #include <linux/ptrace.h>
> +#include <linux/randomize_kstack.h>
> #include <linux/resume_user_mode.h>
> #include <linux/seccomp.h>
> #include <linux/sched.h>
> @@ -150,6 +151,61 @@ static __always_inline long syscall_ente
> }
>
> /**
> + * enter_from_user_mode_randomize_stack - Establish state and add stack randomization
> + * before invoking syscall_enter_from_user_mode_work()
> + * @regs: Pointer to currents pt_regs
> + *
> + * Invoked from architecture specific syscall entry code with interrupts
> + * disabled. The calling code has to be non-instrumentable. When the function
> + * returns all state is correct, interrupts are still disabled and the
> + * subsequent functions can be instrumented.
> + *
> + * Implemented as a macro so that the stack randomization is effective
> + * throughout the function in which it is invoked. An inline would only make it
> + * effective in the scope of the inline function.
> + */
> +#define enter_from_user_mode_randomize_stack(regs) \
> +do { \
> + enter_from_user_mode(regs); \
> + instrumentation_begin(); \
> + add_random_kstack_offset_irqsoff(); \
> + instrumentation_end(); \
> +} while (0)
> +
> +/**
> + * syscall_enter_from_user_mode_randomize_stack - Establish state and check and handle work
> + * before invoking a syscall
> + * @regs: Pointer to currents pt_regs
> + * @syscall: The syscall number
> + *
> + * Invoked from architecture specific syscall entry code with interrupts
> + * disabled. The calling code has to be non-instrumentable. When the
> + * function returns all state is correct, interrupts are enabled and the
> + * subsequent functions can be instrumented.
> + *
> + * This is the combination of enter_from_user_mode_randomize_stack() and
> + * syscall_enter_from_user_mode_work() to be used when there is no
> + * architecture specific work to be done between the two.
> + *
> + * Returns: The original or a modified syscall number. See
> + * syscall_enter_from_user_mode_work() for further explanation.
> + *
> + * Implemented as a macro to make stack randomization effective in the calling
> + * scope.
> + */
> +#define syscall_enter_from_user_mode_randomize_stack(regs, syscall) \
> +({ \
> + enter_from_user_mode_randomize_stack(regs); \
> + \
> + instrumentation_begin(); \
> + local_irq_enable(); \
> + long _ret = syscall_enter_from_user_mode_work(regs, syscall); \
> + instrumentation_end(); \
> + \
> + _ret; \
> +})
Reviewed-by: Jinjie Ruan <ruanjinjie@huawei.com>
> +
> +/**
> * syscall_enter_from_user_mode - Establish state and check and handle work
> * before invoking a syscall
> * @regs: Pointer to currents pt_regs
>
^ 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