linux-kernel - Re: [PATCH] soc: samsung: exynos-pmu: Fix for CONFIG_DEBUG

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <aLshJ11k3c2T-MRs@google.com>
Date: Fri, 5 Sep 2025 17:43:03 +0000
From: Mostafa Saleh <smostafa@...gle.com>
To: Krzysztof Kozlowski <krzk@...nel.org>
Cc: linux-arm-kernel@...ts.infradead.org, linux-samsung-soc@...r.kernel.org,
	linux-kernel@...r.kernel.org, alim.akhtar@...sung.com,
	Peter Griffin <peter.griffin@...aro.org>,
	André Draszik <andre.draszik@...aro.org>
Subject: Re: [PATCH] soc: samsung: exynos-pmu: Fix for CONFIG_DEBUG_PREEMPT

On Fri, Sep 05, 2025 at 07:09:09PM +0200, Krzysztof Kozlowski wrote:
> On 05/09/2025 18:24, Mostafa Saleh wrote:
> > Booting the kernel on Pixel-6 with `CONFIG_DEBUG_PREEMPT` prints the
> > following WARN:
> > 
> > [    0.784187][    T1] BUG: using smp_processor_id() in preemptible [00000000] code: swapper/0/1
> > [    0.784328][    T1] caller is debug_smp_processor_id+0x20/0x30
> > [    0.784433][    T1] CPU: 6 UID: 0 PID: 1 Comm: swapper/0 Not tainted 6.17.0-rc4-gd69eb204c255 #1 PREEMPT
> > [    0.784439][    T1] Hardware name: Oriole (DT)
> > [    0.784441][    T1] Call trace:
> > [    0.784443][    T1]  show_stack+0x34/0xa0 (C)
> > [    0.784453][    T1]  dump_stack_lvl+0x7c/0xb0
> > [    0.784460][    T1]  dump_stack+0x18/0x24
> > [    0.784464][    T1]  check_preemption_disabled+0xf8/0x100
> > [    0.784470][    T1]  debug_smp_processor_id+0x20/0x30
> > [    0.784476][    T1]  gs101_cpuhp_pmu_online+0x40/0x108
> > [    0.784483][    T1]  cpuhp_invoke_callback+0x188/0x2d8
> > [    0.784490][    T1]  cpuhp_issue_call+0xec/0x240
> > [    0.784494][    T1]  __cpuhp_setup_state_cpuslocked+0x140/0x2c0
> > [    0.784499][    T1]  __cpuhp_setup_state+0x58/0x88
> > [    0.784504][    T1]  exynos_pmu_probe+0x2a4/0x380
> > [    0.784508][    T1]  platform_probe+0x64/0xd0
> > [    0.784516][    T1]  really_probe+0xd0/0x3b0
> > [    0.784520][    T1]  __driver_probe_device+0x8c/0x170
> > [    0.784524][    T1]  driver_probe_device+0x44/0x140
> > [    0.784528][    T1]  __device_attach_driver+0xd8/0x180
> > [    0.784532][    T1]  bus_for_each_drv+0x90/0xf8
> > [    0.784536][    T1]  __device_attach+0xa8/0x1d0
> > [    0.784540][    T1]  device_initial_probe+0x1c/0x30
> > [    0.784544][    T1]  bus_probe_device+0xb4/0xc0
> > [    0.784547][    T1]  device_add+0x4d0/0x700
> > [    0.784550][    T1]  of_device_add+0x4c/0x78
> > [    0.784556][    T1]  of_platform_device_create_pdata+0x9c/0x148
> > [    0.784560][    T1]  of_platform_bus_create+0x1d0/0x370
> > [    0.784563][    T1]  of_platform_bus_create+0x234/0x370
> > [    0.784567][    T1]  of_platform_populate+0x84/0x178
> > [    0.784571][    T1]  of_platform_default_populate_init+0xf0/0x120
> > [    0.784579][    T1]  do_one_initcall+0x68/0x2d0
> > [    0.784585][    T1]  kernel_init_freeable+0x2d8/0x358
> > [    0.784589][    T1]  kernel_init+0x28/0x168
> > [    0.784595][    T1]  ret_from_fork+0x10/0x20
> 
> Please trim from unnecessary parts - [time].

Will do.

> 
> > 
> > As this value is only read once, it doesn't require to be stable, so
> 
> Why it does not need to be stable? Onlining wrong CPU number is not
> desired...
> 
> > just use "raw_smp_processor_id" instead.
> 
> You might be just hiding some other real issue, because above stacktrace
> is from gs101_cpuhp_pmu_online() which IRQs disabled and preemption
> disabled. Provide analysis of the warning, instead of just making it
> disappear.

Not sure I understand, how is preemption disabled? that wouldn't fire
in that case.

I am not familiar with this driver, but as I see the value is used
only once, it would be stable, for example using get/put_cpu won't
really matter, because next access doesn't depend on the current CPU.
Otherwise, if you imply that this might not be enough, that means
the driver is already broken even without CONFIG_DEBUG_PREEMP
(which is beyond my knowledge at this point)

> 
> > 
> > Cc: Peter Griffin <peter.griffin@...aro.org>
> > Cc: André Draszik <andre.draszik@...aro.org>
> > 
> 
> No blank lines between tags.
> 
> Missing Fixes tag... and then Cc is not necessary. Please follow
> standard kernel process.

I will add the Fixes.

I am working with Peter and André, so I thought Cc is fine
(according to the process) in:
https://docs.kernel.org/process/5.Posting.html

Thanks,
Mostafa

> 
> > Signed-off-by: Mostafa Saleh <smostafa@...gle.com>
> > ---
> >  drivers/soc/samsung/exynos-pmu.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/soc/samsung/exynos-pmu.c b/drivers/soc/samsung/exynos-pmu.c
> > index a77288f49d24..338f4758a089 100644
> > --- a/drivers/soc/samsung/exynos-pmu.c
> > +++ b/drivers/soc/samsung/exynos-pmu.c
> > @@ -338,7 +338,7 @@ EXPORT_SYMBOL_GPL(exynos_get_pmu_regmap_by_phandle);
> >  
> 
> 
> Best regards,
> Krzysztof