| Message ID | 20250522141410.31315-1-alexghiti@rivosinc.com |
|---|---|
| State | New |
| Headers | show |
| Series | drivers: acpi: Fix platform profile driver on !acpi platforms | expand |
On Thu, May 22, 2025, at 16:13, Alexandre Ghiti wrote: > Fixes: 77be5cacb2c2 ("ACPI: platform_profile: Create class for ACPI > platform profile") > Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com> Reviewed-by: Arnd Bergmann <arnd@arndb.de>
On Thu, May 22, 2025, at 10:13 AM, Alexandre Ghiti wrote: > The platform profile driver is loaded even on platforms that do not have > acpi enabled. The initialization of the sysfs entries was recently moved > from platform_profile_register() to the module init call, and those > entries need acpi_kobj to be initialized which is not the case when acpi > is disabled. > > This results in the following warning: > > WARNING: CPU: 5 PID: 1 at fs/sysfs/group.c:131 > internal_create_group+0xa22/0xdd8 > Modules linked in: > CPU: 5 UID: 0 PID: 1 Comm: swapper/0 Tainted: G W > 6.15.0-rc7-dirty #6 PREEMPT > Tainted: [W]=WARN > Hardware name: riscv-virtio,qemu (DT) > epc : internal_create_group+0xa22/0xdd8 > ra : internal_create_group+0xa22/0xdd8 > > Call Trace: > > internal_create_group+0xa22/0xdd8 > sysfs_create_group+0x22/0x2e > platform_profile_init+0x74/0xb2 > do_one_initcall+0x198/0xa9e > kernel_init_freeable+0x6d8/0x780 > kernel_init+0x28/0x24c > ret_from_fork+0xe/0x18 > > Fix this by checking if acpi is enabled before trying to create sysfs > entries. > > Fixes: 77be5cacb2c2 ("ACPI: platform_profile: Create class for ACPI > platform profile") > Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com> > --- > drivers/acpi/platform_profile.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/acpi/platform_profile.c b/drivers/acpi/platform_profile.c > index ffbfd32f4cf1..b43f4459a4f6 100644 > --- a/drivers/acpi/platform_profile.c > +++ b/drivers/acpi/platform_profile.c > @@ -688,6 +688,9 @@ static int __init platform_profile_init(void) > { > int err; > > + if (acpi_disabled) > + return -EOPNOTSUPP; > + > err = class_register(&platform_profile_class); > if (err) > return err; > -- > 2.43.0 Looks good to me. Reviewed-by: Mark Pearson <mpearson-lenovo@squebb.ca> Mark
Am 22.05.25 um 16:13 schrieb Alexandre Ghiti: > The platform profile driver is loaded even on platforms that do not have > acpi enabled. The initialization of the sysfs entries was recently moved > from platform_profile_register() to the module init call, and those > entries need acpi_kobj to be initialized which is not the case when acpi > is disabled. > > This results in the following warning: > > WARNING: CPU: 5 PID: 1 at fs/sysfs/group.c:131 internal_create_group+0xa22/0xdd8 > Modules linked in: > CPU: 5 UID: 0 PID: 1 Comm: swapper/0 Tainted: G W 6.15.0-rc7-dirty #6 PREEMPT > Tainted: [W]=WARN > Hardware name: riscv-virtio,qemu (DT) > epc : internal_create_group+0xa22/0xdd8 > ra : internal_create_group+0xa22/0xdd8 > > Call Trace: > > internal_create_group+0xa22/0xdd8 > sysfs_create_group+0x22/0x2e > platform_profile_init+0x74/0xb2 > do_one_initcall+0x198/0xa9e > kernel_init_freeable+0x6d8/0x780 > kernel_init+0x28/0x24c > ret_from_fork+0xe/0x18 > > Fix this by checking if acpi is enabled before trying to create sysfs > entries. I already submitted a patch for this problem (see https://lore.kernel.org/linux-acpi/a6d92cdd-4dc3-4080-9ed9-5b1f02f247e0@gmx.de/T/) that only disables the legacy sysfs interface while keeping the class-based interface functional as it does not depend on ACPI at all. Thank, Armin Wolf > Fixes: 77be5cacb2c2 ("ACPI: platform_profile: Create class for ACPI platform profile") > Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com> > --- > drivers/acpi/platform_profile.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/acpi/platform_profile.c b/drivers/acpi/platform_profile.c > index ffbfd32f4cf1..b43f4459a4f6 100644 > --- a/drivers/acpi/platform_profile.c > +++ b/drivers/acpi/platform_profile.c > @@ -688,6 +688,9 @@ static int __init platform_profile_init(void) > { > int err; > > + if (acpi_disabled) > + return -EOPNOTSUPP; > + > err = class_register(&platform_profile_class); > if (err) > return err;
On 5/22/25 22:04, Armin Wolf wrote: > Am 22.05.25 um 16:13 schrieb Alexandre Ghiti: > >> The platform profile driver is loaded even on platforms that do not have >> acpi enabled. The initialization of the sysfs entries was recently moved >> from platform_profile_register() to the module init call, and those >> entries need acpi_kobj to be initialized which is not the case when acpi >> is disabled. >> >> This results in the following warning: >> >> WARNING: CPU: 5 PID: 1 at fs/sysfs/group.c:131 >> internal_create_group+0xa22/0xdd8 >> Modules linked in: >> CPU: 5 UID: 0 PID: 1 Comm: swapper/0 Tainted: G W >> 6.15.0-rc7-dirty #6 PREEMPT >> Tainted: [W]=WARN >> Hardware name: riscv-virtio,qemu (DT) >> epc : internal_create_group+0xa22/0xdd8 >> ra : internal_create_group+0xa22/0xdd8 >> >> Call Trace: >> >> internal_create_group+0xa22/0xdd8 >> sysfs_create_group+0x22/0x2e >> platform_profile_init+0x74/0xb2 >> do_one_initcall+0x198/0xa9e >> kernel_init_freeable+0x6d8/0x780 >> kernel_init+0x28/0x24c >> ret_from_fork+0xe/0x18 >> >> Fix this by checking if acpi is enabled before trying to create sysfs >> entries. > > I already submitted a patch for this problem (see > https://lore.kernel.org/linux-acpi/a6d92cdd-4dc3-4080-9ed9-5b1f02f247e0@gmx.de/T/) > that only disables the legacy sysfs interface while keeping the > class-based interface functional > as it does not depend on ACPI at all. Great, I understand if your patchset is not merged for rc1 but it would be nice to have it merged in 6.16 though to fix riscv syzkaller instance. Perhaps you could add the Fixes tag that Arnd mentioned too? Thanks, Alex > > Thank, > Armin Wolf > >> Fixes: 77be5cacb2c2 ("ACPI: platform_profile: Create class for ACPI >> platform profile") >> Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com> >> --- >> drivers/acpi/platform_profile.c | 3 +++ >> 1 file changed, 3 insertions(+) >> >> diff --git a/drivers/acpi/platform_profile.c >> b/drivers/acpi/platform_profile.c >> index ffbfd32f4cf1..b43f4459a4f6 100644 >> --- a/drivers/acpi/platform_profile.c >> +++ b/drivers/acpi/platform_profile.c >> @@ -688,6 +688,9 @@ static int __init platform_profile_init(void) >> { >> int err; >> + if (acpi_disabled) >> + return -EOPNOTSUPP; >> + >> err = class_register(&platform_profile_class); >> if (err) >> return err; > > _______________________________________________ > linux-riscv mailing list > linux-riscv@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/linux-riscv
On Thu, May 22, 2025, at 22:04, Armin Wolf wrote: > Am 22.05.25 um 16:13 schrieb Alexandre Ghiti: > > I already submitted a patch for this problem (see > https://lore.kernel.org/linux-acpi/a6d92cdd-4dc3-4080-9ed9-5b1f02f247e0@gmx.de/T/) > that only disables the legacy sysfs interface while keeping the > class-based interface functional > as it does not depend on ACPI at all. Just for my understanding: what users of the platform profile are there that work without ACPI? I see that CONFIG_ACPI_PLATFORM_PROFILE is hidden under CONFIG_ACPI and cannot be selected in configurations that turn off ACPI, so if that is an intended usecase, there is probably still something wrong in Kconfig. Should the driver be moved out of drivers/acpi to drivers/platform in order to let non-ACPI platforms use it? Arnd
On Fri, May 23, 2025 at 12:11 PM Alexandre Ghiti <alex@ghiti.fr> wrote: > > On 5/22/25 22:04, Armin Wolf wrote: > > Am 22.05.25 um 16:13 schrieb Alexandre Ghiti: > > > >> The platform profile driver is loaded even on platforms that do not have > >> acpi enabled. The initialization of the sysfs entries was recently moved > >> from platform_profile_register() to the module init call, and those > >> entries need acpi_kobj to be initialized which is not the case when acpi > >> is disabled. > >> > >> This results in the following warning: > >> > >> WARNING: CPU: 5 PID: 1 at fs/sysfs/group.c:131 > >> internal_create_group+0xa22/0xdd8 > >> Modules linked in: > >> CPU: 5 UID: 0 PID: 1 Comm: swapper/0 Tainted: G W > >> 6.15.0-rc7-dirty #6 PREEMPT > >> Tainted: [W]=WARN > >> Hardware name: riscv-virtio,qemu (DT) > >> epc : internal_create_group+0xa22/0xdd8 > >> ra : internal_create_group+0xa22/0xdd8 > >> > >> Call Trace: > >> > >> internal_create_group+0xa22/0xdd8 > >> sysfs_create_group+0x22/0x2e > >> platform_profile_init+0x74/0xb2 > >> do_one_initcall+0x198/0xa9e > >> kernel_init_freeable+0x6d8/0x780 > >> kernel_init+0x28/0x24c > >> ret_from_fork+0xe/0x18 > >> > >> Fix this by checking if acpi is enabled before trying to create sysfs > >> entries. > > > > I already submitted a patch for this problem (see > > https://lore.kernel.org/linux-acpi/a6d92cdd-4dc3-4080-9ed9-5b1f02f247e0@gmx.de/T/) > > that only disables the legacy sysfs interface while keeping the > > class-based interface functional > > as it does not depend on ACPI at all. > > > Great, I understand if your patchset is not merged for rc1 but it would > be nice to have it merged in 6.16 though to fix riscv syzkaller > instance. Perhaps you could add the Fixes tag that Arnd mentioned too? I actually prefer your patch to the Armin's one because there are questions regarding the latter (see the most recent message from Arnd in this thread). Thanks! > >> Fixes: 77be5cacb2c2 ("ACPI: platform_profile: Create class for ACPI > >> platform profile") > >> Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com> > >> --- > >> drivers/acpi/platform_profile.c | 3 +++ > >> 1 file changed, 3 insertions(+) > >> > >> diff --git a/drivers/acpi/platform_profile.c > >> b/drivers/acpi/platform_profile.c > >> index ffbfd32f4cf1..b43f4459a4f6 100644 > >> --- a/drivers/acpi/platform_profile.c > >> +++ b/drivers/acpi/platform_profile.c > >> @@ -688,6 +688,9 @@ static int __init platform_profile_init(void) > >> { > >> int err; > >> + if (acpi_disabled) > >> + return -EOPNOTSUPP; > >> + > >> err = class_register(&platform_profile_class); > >> if (err) > >> return err; > > > > _______________________________________________ > > linux-riscv mailing list > > linux-riscv@lists.infradead.org > > http://lists.infradead.org/mailman/listinfo/linux-riscv
On Fri, May 23, 2025 at 12:50 PM Rafael J. Wysocki <rafael@kernel.org> wrote: > > On Fri, May 23, 2025 at 12:11 PM Alexandre Ghiti <alex@ghiti.fr> wrote: > > > > On 5/22/25 22:04, Armin Wolf wrote: > > > Am 22.05.25 um 16:13 schrieb Alexandre Ghiti: > > > > > >> The platform profile driver is loaded even on platforms that do not have > > >> acpi enabled. The initialization of the sysfs entries was recently moved > > >> from platform_profile_register() to the module init call, and those > > >> entries need acpi_kobj to be initialized which is not the case when acpi > > >> is disabled. > > >> > > >> This results in the following warning: > > >> > > >> WARNING: CPU: 5 PID: 1 at fs/sysfs/group.c:131 > > >> internal_create_group+0xa22/0xdd8 > > >> Modules linked in: > > >> CPU: 5 UID: 0 PID: 1 Comm: swapper/0 Tainted: G W > > >> 6.15.0-rc7-dirty #6 PREEMPT > > >> Tainted: [W]=WARN > > >> Hardware name: riscv-virtio,qemu (DT) > > >> epc : internal_create_group+0xa22/0xdd8 > > >> ra : internal_create_group+0xa22/0xdd8 > > >> > > >> Call Trace: > > >> > > >> internal_create_group+0xa22/0xdd8 > > >> sysfs_create_group+0x22/0x2e > > >> platform_profile_init+0x74/0xb2 > > >> do_one_initcall+0x198/0xa9e > > >> kernel_init_freeable+0x6d8/0x780 > > >> kernel_init+0x28/0x24c > > >> ret_from_fork+0xe/0x18 > > >> > > >> Fix this by checking if acpi is enabled before trying to create sysfs > > >> entries. > > > > > > I already submitted a patch for this problem (see > > > https://lore.kernel.org/linux-acpi/a6d92cdd-4dc3-4080-9ed9-5b1f02f247e0@gmx.de/T/) > > > that only disables the legacy sysfs interface while keeping the > > > class-based interface functional > > > as it does not depend on ACPI at all. > > > > > > Great, I understand if your patchset is not merged for rc1 but it would > > be nice to have it merged in 6.16 though to fix riscv syzkaller > > instance. Perhaps you could add the Fixes tag that Arnd mentioned too? > > I actually prefer your patch to the Armin's one because there are > questions regarding the latter (see the most recent message from Arnd > in this thread). And so it has been applied as 6.16 material now, thanks! > > >> Fixes: 77be5cacb2c2 ("ACPI: platform_profile: Create class for ACPI > > >> platform profile") > > >> Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com> > > >> --- > > >> drivers/acpi/platform_profile.c | 3 +++ > > >> 1 file changed, 3 insertions(+) > > >> > > >> diff --git a/drivers/acpi/platform_profile.c > > >> b/drivers/acpi/platform_profile.c > > >> index ffbfd32f4cf1..b43f4459a4f6 100644 > > >> --- a/drivers/acpi/platform_profile.c > > >> +++ b/drivers/acpi/platform_profile.c > > >> @@ -688,6 +688,9 @@ static int __init platform_profile_init(void) > > >> { > > >> int err; > > >> + if (acpi_disabled) > > >> + return -EOPNOTSUPP; > > >> + > > >> err = class_register(&platform_profile_class); > > >> if (err) > > >> return err; > > > > > > _______________________________________________ > > > linux-riscv mailing list > > > linux-riscv@lists.infradead.org > > > http://lists.infradead.org/mailman/listinfo/linux-riscv
Hi Rafael, On 5/23/25 16:46, Rafael J. Wysocki wrote: > On Fri, May 23, 2025 at 12:50 PM Rafael J. Wysocki <rafael@kernel.org> wrote: >> On Fri, May 23, 2025 at 12:11 PM Alexandre Ghiti <alex@ghiti.fr> wrote: >>> On 5/22/25 22:04, Armin Wolf wrote: >>>> Am 22.05.25 um 16:13 schrieb Alexandre Ghiti: >>>> >>>>> The platform profile driver is loaded even on platforms that do not have >>>>> acpi enabled. The initialization of the sysfs entries was recently moved >>>>> from platform_profile_register() to the module init call, and those >>>>> entries need acpi_kobj to be initialized which is not the case when acpi >>>>> is disabled. >>>>> >>>>> This results in the following warning: >>>>> >>>>> WARNING: CPU: 5 PID: 1 at fs/sysfs/group.c:131 >>>>> internal_create_group+0xa22/0xdd8 >>>>> Modules linked in: >>>>> CPU: 5 UID: 0 PID: 1 Comm: swapper/0 Tainted: G W >>>>> 6.15.0-rc7-dirty #6 PREEMPT >>>>> Tainted: [W]=WARN >>>>> Hardware name: riscv-virtio,qemu (DT) >>>>> epc : internal_create_group+0xa22/0xdd8 >>>>> ra : internal_create_group+0xa22/0xdd8 >>>>> >>>>> Call Trace: >>>>> >>>>> internal_create_group+0xa22/0xdd8 >>>>> sysfs_create_group+0x22/0x2e >>>>> platform_profile_init+0x74/0xb2 >>>>> do_one_initcall+0x198/0xa9e >>>>> kernel_init_freeable+0x6d8/0x780 >>>>> kernel_init+0x28/0x24c >>>>> ret_from_fork+0xe/0x18 >>>>> >>>>> Fix this by checking if acpi is enabled before trying to create sysfs >>>>> entries. >>>> I already submitted a patch for this problem (see >>>> https://lore.kernel.org/linux-acpi/a6d92cdd-4dc3-4080-9ed9-5b1f02f247e0@gmx.de/T/) >>>> that only disables the legacy sysfs interface while keeping the >>>> class-based interface functional >>>> as it does not depend on ACPI at all. >>> >>> Great, I understand if your patchset is not merged for rc1 but it would >>> be nice to have it merged in 6.16 though to fix riscv syzkaller >>> instance. Perhaps you could add the Fixes tag that Arnd mentioned too? >> I actually prefer your patch to the Armin's one because there are >> questions regarding the latter (see the most recent message from Arnd >> in this thread). > And so it has been applied as 6.16 material now, thanks! Thank you very much! Alex > >>>>> Fixes: 77be5cacb2c2 ("ACPI: platform_profile: Create class for ACPI >>>>> platform profile") >>>>> Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com> >>>>> --- >>>>> drivers/acpi/platform_profile.c | 3 +++ >>>>> 1 file changed, 3 insertions(+) >>>>> >>>>> diff --git a/drivers/acpi/platform_profile.c >>>>> b/drivers/acpi/platform_profile.c >>>>> index ffbfd32f4cf1..b43f4459a4f6 100644 >>>>> --- a/drivers/acpi/platform_profile.c >>>>> +++ b/drivers/acpi/platform_profile.c >>>>> @@ -688,6 +688,9 @@ static int __init platform_profile_init(void) >>>>> { >>>>> int err; >>>>> + if (acpi_disabled) >>>>> + return -EOPNOTSUPP; >>>>> + >>>>> err = class_register(&platform_profile_class); >>>>> if (err) >>>>> return err; >>>> _______________________________________________ >>>> linux-riscv mailing list >>>> linux-riscv@lists.infradead.org >>>> http://lists.infradead.org/mailman/listinfo/linux-riscv > _______________________________________________ > linux-riscv mailing list > linux-riscv@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/linux-riscv
Hello: This patch was applied to riscv/linux.git (fixes) by Rafael J. Wysocki <rafael.j.wysocki@intel.com>: On Thu, 22 May 2025 16:13:56 +0200 you wrote: > The platform profile driver is loaded even on platforms that do not have > acpi enabled. The initialization of the sysfs entries was recently moved > from platform_profile_register() to the module init call, and those > entries need acpi_kobj to be initialized which is not the case when acpi > is disabled. > > This results in the following warning: > > [...] Here is the summary with links: - drivers: acpi: Fix platform profile driver on !acpi platforms https://git.kernel.org/riscv/c/dd133162c9cf You are awesome, thank you!
diff --git a/drivers/acpi/platform_profile.c b/drivers/acpi/platform_profile.c index ffbfd32f4cf1..b43f4459a4f6 100644 --- a/drivers/acpi/platform_profile.c +++ b/drivers/acpi/platform_profile.c @@ -688,6 +688,9 @@ static int __init platform_profile_init(void) { int err; + if (acpi_disabled) + return -EOPNOTSUPP; + err = class_register(&platform_profile_class); if (err) return err;
The platform profile driver is loaded even on platforms that do not have acpi enabled. The initialization of the sysfs entries was recently moved from platform_profile_register() to the module init call, and those entries need acpi_kobj to be initialized which is not the case when acpi is disabled. This results in the following warning: WARNING: CPU: 5 PID: 1 at fs/sysfs/group.c:131 internal_create_group+0xa22/0xdd8 Modules linked in: CPU: 5 UID: 0 PID: 1 Comm: swapper/0 Tainted: G W 6.15.0-rc7-dirty #6 PREEMPT Tainted: [W]=WARN Hardware name: riscv-virtio,qemu (DT) epc : internal_create_group+0xa22/0xdd8 ra : internal_create_group+0xa22/0xdd8 Call Trace: internal_create_group+0xa22/0xdd8 sysfs_create_group+0x22/0x2e platform_profile_init+0x74/0xb2 do_one_initcall+0x198/0xa9e kernel_init_freeable+0x6d8/0x780 kernel_init+0x28/0x24c ret_from_fork+0xe/0x18 Fix this by checking if acpi is enabled before trying to create sysfs entries. Fixes: 77be5cacb2c2 ("ACPI: platform_profile: Create class for ACPI platform profile") Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com> --- drivers/acpi/platform_profile.c | 3 +++ 1 file changed, 3 insertions(+)