[rt-devel:linux-6.10.y-rt-rebase 63/93] kernel/printk/printk.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction

[rt-devel:linux-6.10.y-rt-rebase 63/93] kernel/printk/printk.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction

[kernel test robot]

tree:   https://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-rt-devel.git linux-6.10.y-rt-rebase
head:   6d20efa57a896036453f4b061ae22601ef61b494
commit: 127f3e5b794bd6fa90d4649318dd898455efa6d7 [63/93] printk: Provide threadprintk boot argument
config: loongarch-defconfig (https://download.01.org/0day-ci/archive/20240530/202405300545.mVqyyvyr-lkp@intel.com/config)
compiler: loongarch64-linux-gcc (GCC) 13.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240530/202405300545.mVqyyvyr-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202405300545.mVqyyvyr-lkp@intel.com/

All warnings (new ones prefixed by >>):

>> kernel/printk/printk.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction
--
>> kernel/printk/nbcon.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction


objdump-func vmlinux.o __jump_table:

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

Re: [rt-devel:linux-6.10.y-rt-rebase 63/93] kernel/printk/printk.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction

[John Ogness]

On 2024-05-30, kernel test robot <lkp@intel.com> wrote:
> tree:   https://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-rt-devel.git linux-6.10.y-rt-rebase
> head:   6d20efa57a896036453f4b061ae22601ef61b494
> commit: 127f3e5b794bd6fa90d4649318dd898455efa6d7 [63/93] printk: Provide threadprintk boot argument
> config: loongarch-defconfig (https://download.01.org/0day-ci/archive/20240530/202405300545.mVqyyvyr-lkp@intel.com/config)
> compiler: loongarch64-linux-gcc (GCC) 13.2.0
> reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240530/202405300545.mVqyyvyr-lkp@intel.com/reproduce)
>
> If you fix the issue in a separate patch/commit (i.e. not just a new version of
> the same patch/commit), kindly add following tags
> | Reported-by: kernel test robot <lkp@intel.com>
> | Closes: https://lore.kernel.org/oe-kbuild-all/202405300545.mVqyyvyr-lkp@intel.com/
>
> All warnings (new ones prefixed by >>):
>
>>> kernel/printk/printk.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction
> --
>>> kernel/printk/nbcon.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction

On loongarch, it looks like all users of the jump_table are causing this
warning. I will assume this is something that needs to be fixed at the
arch level (either in the kernel or the compiler).

John Ogness

Re: [rt-devel:linux-6.10.y-rt-rebase 63/93] kernel/printk/printk.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction

[Yujie Liu]

+ Loongarch folks

Hi Huacai and Xuerui,

Could you please help take a look at this case? Thanks.

On Mon, Jun 03, 2024 at 11:48:55AM +0206, John Ogness wrote:
> On 2024-05-30, kernel test robot <lkp@intel.com> wrote:
> > tree:   https://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-rt-devel.git linux-6.10.y-rt-rebase
> > head:   6d20efa57a896036453f4b061ae22601ef61b494
> > commit: 127f3e5b794bd6fa90d4649318dd898455efa6d7 [63/93] printk: Provide threadprintk boot argument
> > config: loongarch-defconfig (https://download.01.org/0day-ci/archive/20240530/202405300545.mVqyyvyr-lkp@intel.com/config)
> > compiler: loongarch64-linux-gcc (GCC) 13.2.0
> > reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240530/202405300545.mVqyyvyr-lkp@intel.com/reproduce)
> >
> > If you fix the issue in a separate patch/commit (i.e. not just a new version of
> > the same patch/commit), kindly add following tags
> > | Reported-by: kernel test robot <lkp@intel.com>
> > | Closes: https://lore.kernel.org/oe-kbuild-all/202405300545.mVqyyvyr-lkp@intel.com/
> >
> > All warnings (new ones prefixed by >>):
> >
> >>> kernel/printk/printk.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction
> > --
> >>> kernel/printk/nbcon.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction
> 
> On loongarch, it looks like all users of the jump_table are causing this
> warning. I will assume this is something that needs to be fixed at the
> arch level (either in the kernel or the compiler).
> 
> John Ogness
> 

Re: [rt-devel:linux-6.10.y-rt-rebase 63/93] kernel/printk/printk.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction

[Huacai Chen]

On Tue, Jun 4, 2024 at 9:07 AM Yujie Liu <yujie.liu@intel.com> wrote:
>
> + Loongarch folks
>
> Hi Huacai and Xuerui,
>
> Could you please help take a look at this case? Thanks.
We are attempting to solve this problem, please wait some time.

Huacai

>
> On Mon, Jun 03, 2024 at 11:48:55AM +0206, John Ogness wrote:
> > On 2024-05-30, kernel test robot <lkp@intel.com> wrote:
> > > tree:   https://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-rt-devel.git linux-6.10.y-rt-rebase
> > > head:   6d20efa57a896036453f4b061ae22601ef61b494
> > > commit: 127f3e5b794bd6fa90d4649318dd898455efa6d7 [63/93] printk: Provide threadprintk boot argument
> > > config: loongarch-defconfig (https://download.01.org/0day-ci/archive/20240530/202405300545.mVqyyvyr-lkp@intel.com/config)
> > > compiler: loongarch64-linux-gcc (GCC) 13.2.0
> > > reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240530/202405300545.mVqyyvyr-lkp@intel.com/reproduce)
> > >
> > > If you fix the issue in a separate patch/commit (i.e. not just a new version of
> > > the same patch/commit), kindly add following tags
> > > | Reported-by: kernel test robot <lkp@intel.com>
> > > | Closes: https://lore.kernel.org/oe-kbuild-all/202405300545.mVqyyvyr-lkp@intel.com/
> > >
> > > All warnings (new ones prefixed by >>):
> > >
> > >>> kernel/printk/printk.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction
> > > --
> > >>> kernel/printk/nbcon.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction
> >
> > On loongarch, it looks like all users of the jump_table are causing this
> > warning. I will assume this is something that needs to be fixed at the
> > arch level (either in the kernel or the compiler).
> >
> > John Ogness
> >

Re: [rt-devel:linux-6.10.y-rt-rebase 63/93] kernel/printk/printk.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction

[Tiezhu Yang]

On 06/04/2024 08:59 AM, Yujie Liu wrote:
> + Loongarch folks
>
> Hi Huacai and Xuerui,
>
> Could you please help take a look at this case? Thanks.
>
> On Mon, Jun 03, 2024 at 11:48:55AM +0206, John Ogness wrote:
>> On 2024-05-30, kernel test robot <lkp@intel.com> wrote:
>>> tree:   https://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-rt-devel.git linux-6.10.y-rt-rebase
>>> head:   6d20efa57a896036453f4b061ae22601ef61b494
>>> commit: 127f3e5b794bd6fa90d4649318dd898455efa6d7 [63/93] printk: Provide threadprintk boot argument
>>> config: loongarch-defconfig (https://download.01.org/0day-ci/archive/20240530/202405300545.mVqyyvyr-lkp@intel.com/config)
>>> compiler: loongarch64-linux-gcc (GCC) 13.2.0
>>> reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240530/202405300545.mVqyyvyr-lkp@intel.com/reproduce)
>>>
>>> If you fix the issue in a separate patch/commit (i.e. not just a new version of
>>> the same patch/commit), kindly add following tags
>>> | Reported-by: kernel test robot <lkp@intel.com>
>>> | Closes: https://lore.kernel.org/oe-kbuild-all/202405300545.mVqyyvyr-lkp@intel.com/
>>>
>>> All warnings (new ones prefixed by >>):
>>>
>>>>> kernel/printk/printk.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction
>>> --
>>>>> kernel/printk/nbcon.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction
>>
>> On loongarch, it looks like all users of the jump_table are causing this
>> warning. I will assume this is something that needs to be fixed at the
>> arch level (either in the kernel or the compiler).
>>
>> John Ogness
>>

Here are my thoughts, anyway it is up to the maintainers.

After commit 9cc1df421f00 ("LoongArch: Update Loongson-3 default
config file"), there exist some build regressions about objtool
warnings on LoongArch. This is because objtool does not work well
with jump tables, -fno-jump-tables is already added in Makefile,
but there are some special cases, some warnings can be reproduced
because the code has __annotate_jump_table, some warnings can not
be reproduced with unknown reason (maybe due to the lower version
of compiler toolchain). In order to avoid the warnings reported
by kernel test robot, it is better to remove CONFIG_UNWINDER_ORC=y
in defconfig for now, the proper time to enable ORC stack unwinder 
should after objtool supports jump tables on LoongArch.

arch/loongarch/Makefile

   ifdef CONFIG_OBJTOOL
   KBUILD_CFLAGS           += -fno-jump-tables
   endif

include/linux/compiler.h

   /* Annotate a C jump table to allow objtool to follow the code flow */
   #define __annotate_jump_table __section(".rodata..c_jump_table")

Like this:

diff --git a/arch/loongarch/configs/loongson3_defconfig 
b/arch/loongarch/configs/loongson3_defconfig
index b4252c357c8e..0eb14e9259c2 100644
--- a/arch/loongarch/configs/loongson3_defconfig
+++ b/arch/loongarch/configs/loongson3_defconfig
@@ -983,4 +983,3 @@ CONFIG_DEBUG_FS=y
  CONFIG_SCHEDSTATS=y
  # CONFIG_DEBUG_PREEMPT is not set
  # CONFIG_FTRACE is not set
-CONFIG_UNWINDER_ORC=y

Thanks,
Tiezhu

Re: [rt-devel:linux-6.10.y-rt-rebase 63/93] kernel/printk/printk.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction

[Sebastian Andrzej Siewior]

On 2024-06-04 17:49:35 [+0800], Tiezhu Yang wrote:
> Like this:
> 
> diff --git a/arch/loongarch/configs/loongson3_defconfig
> b/arch/loongarch/configs/loongson3_defconfig
> index b4252c357c8e..0eb14e9259c2 100644
> --- a/arch/loongarch/configs/loongson3_defconfig
> +++ b/arch/loongarch/configs/loongson3_defconfig
> @@ -983,4 +983,3 @@ CONFIG_DEBUG_FS=y
>  CONFIG_SCHEDSTATS=y
>  # CONFIG_DEBUG_PREEMPT is not set
>  # CONFIG_FTRACE is not set
> -CONFIG_UNWINDER_ORC=y

If ORC/objtool is broken on loongarch, shouldn't you disable it then or
mask at least via

diff --git a/arch/loongarch/Kconfig.debug b/arch/loongarch/Kconfig.debug
index 98d60630c3d4b..2bce03b8f8309 100644
--- a/arch/loongarch/Kconfig.debug
+++ b/arch/loongarch/Kconfig.debug
@@ -28,6 +28,7 @@ config UNWINDER_PROLOGUE
 
 config UNWINDER_ORC
 	bool "ORC unwinder"
+	depends on BROKEN
 	select OBJTOOL
 	help
 	  This option enables the ORC (Oops Rewind Capability) unwinder for

instead? Removing it from defconfig makes it still available to to
randconfig.

> Thanks,
> Tiezhu

Sebastian

Re: [rt-devel:linux-6.10.y-rt-rebase 63/93] kernel/printk/printk.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction

[Tiezhu Yang]

On 06/04/2024 06:03 PM, Sebastian Andrzej Siewior wrote:
> On 2024-06-04 17:49:35 [+0800], Tiezhu Yang wrote:
>> Like this:
>>
>> diff --git a/arch/loongarch/configs/loongson3_defconfig
>> b/arch/loongarch/configs/loongson3_defconfig
>> index b4252c357c8e..0eb14e9259c2 100644
>> --- a/arch/loongarch/configs/loongson3_defconfig
>> +++ b/arch/loongarch/configs/loongson3_defconfig
>> @@ -983,4 +983,3 @@ CONFIG_DEBUG_FS=y
>>  CONFIG_SCHEDSTATS=y
>>  # CONFIG_DEBUG_PREEMPT is not set
>>  # CONFIG_FTRACE is not set
>> -CONFIG_UNWINDER_ORC=y
>
> If ORC/objtool is broken on loongarch, shouldn't you disable it then or
> mask at least via
>
> diff --git a/arch/loongarch/Kconfig.debug b/arch/loongarch/Kconfig.debug
> index 98d60630c3d4b..2bce03b8f8309 100644
> --- a/arch/loongarch/Kconfig.debug
> +++ b/arch/loongarch/Kconfig.debug
> @@ -28,6 +28,7 @@ config UNWINDER_PROLOGUE
>
>  config UNWINDER_ORC
>  	bool "ORC unwinder"
> +	depends on BROKEN
>  	select OBJTOOL
>  	help
>  	  This option enables the ORC (Oops Rewind Capability) unwinder for
>
> instead? Removing it from defconfig makes it still available to to
> randconfig.

Yes, good suggestion, I think so.
Let us wait for more comments.

Thanks,
Tiezhu

Re: [rt-devel:linux-6.10.y-rt-rebase 63/93] kernel/printk/printk.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction

[Xi Ruoyao]

On Tue, 2024-06-04 at 12:03 +0200, Sebastian Andrzej Siewior wrote:
> On 2024-06-04 17:49:35 [+0800], Tiezhu Yang wrote:
> > Like this:
> > 
> > diff --git a/arch/loongarch/configs/loongson3_defconfig
> > b/arch/loongarch/configs/loongson3_defconfig
> > index b4252c357c8e..0eb14e9259c2 100644
> > --- a/arch/loongarch/configs/loongson3_defconfig
> > +++ b/arch/loongarch/configs/loongson3_defconfig
> > @@ -983,4 +983,3 @@ CONFIG_DEBUG_FS=y
> >  CONFIG_SCHEDSTATS=y
> >  # CONFIG_DEBUG_PREEMPT is not set
> >  # CONFIG_FTRACE is not set
> > -CONFIG_UNWINDER_ORC=y
> 
> If ORC/objtool is broken on loongarch, shouldn't you disable it then or
> mask at least via
> 
> diff --git a/arch/loongarch/Kconfig.debug b/arch/loongarch/Kconfig.debug
> index 98d60630c3d4b..2bce03b8f8309 100644
> --- a/arch/loongarch/Kconfig.debug
> +++ b/arch/loongarch/Kconfig.debug
> @@ -28,6 +28,7 @@ config UNWINDER_PROLOGUE
>  
>  config UNWINDER_ORC
>  	bool "ORC unwinder"
> +	depends on BROKEN
>  	select OBJTOOL
>  	help
>  	  This option enables the ORC (Oops Rewind Capability) unwinder for
> 
> instead? Removing it from defconfig makes it still available to to
> randconfig.

I agree but...

diff --git a/arch/loongarch/Kconfig b/arch/loongarch/Kconfig
index e38139c576ee..4706aa948a5f 100644
--- a/arch/loongarch/Kconfig
+++ b/arch/loongarch/Kconfig
@@ -143,7 +143,7 @@ config LOONGARCH
        select HAVE_LIVEPATCH
        select HAVE_MOD_ARCH_SPECIFIC
        select HAVE_NMI
-       select HAVE_OBJTOOL if AS_HAS_EXPLICIT_RELOCS
+       select HAVE_OBJTOOL if AS_HAS_EXPLICIT_RELOCS && BROKEN
        select HAVE_PCI
        select HAVE_PERF_EVENTS
        select HAVE_PERF_REGS

should be better?  It looks like everything depending on objtool is
broken, not only ORC.

-- 
Xi Ruoyao <xry111@xry111.site>
School of Aerospace Science and Technology, Xidian University

Re: [rt-devel:linux-6.10.y-rt-rebase 63/93] kernel/printk/printk.o: warning: objtool: __jump_table+0x0: special: can't find orig instruction

[Xi Ruoyao]

On Tue, 2024-06-04 at 20:09 +0800, Xi Ruoyao wrote:
> On Tue, 2024-06-04 at 12:03 +0200, Sebastian Andrzej Siewior wrote:
> > On 2024-06-04 17:49:35 [+0800], Tiezhu Yang wrote:
> > > Like this:
> > > 
> > > diff --git a/arch/loongarch/configs/loongson3_defconfig
> > > b/arch/loongarch/configs/loongson3_defconfig
> > > index b4252c357c8e..0eb14e9259c2 100644
> > > --- a/arch/loongarch/configs/loongson3_defconfig
> > > +++ b/arch/loongarch/configs/loongson3_defconfig
> > > @@ -983,4 +983,3 @@ CONFIG_DEBUG_FS=y
> > >  CONFIG_SCHEDSTATS=y
> > >  # CONFIG_DEBUG_PREEMPT is not set
> > >  # CONFIG_FTRACE is not set
> > > -CONFIG_UNWINDER_ORC=y
> > 
> > If ORC/objtool is broken on loongarch, shouldn't you disable it then or
> > mask at least via
> > 
> > diff --git a/arch/loongarch/Kconfig.debug b/arch/loongarch/Kconfig.debug
> > index 98d60630c3d4b..2bce03b8f8309 100644
> > --- a/arch/loongarch/Kconfig.debug
> > +++ b/arch/loongarch/Kconfig.debug
> > @@ -28,6 +28,7 @@ config UNWINDER_PROLOGUE
> >  
> >  config UNWINDER_ORC
> >  	bool "ORC unwinder"
> > +	depends on BROKEN
> >  	select OBJTOOL
> >  	help
> >  	  This option enables the ORC (Oops Rewind Capability) unwinder for
> > 
> > instead? Removing it from defconfig makes it still available to to
> > randconfig.
> 
> I agree but...
> 
> diff --git a/arch/loongarch/Kconfig b/arch/loongarch/Kconfig
> index e38139c576ee..4706aa948a5f 100644
> --- a/arch/loongarch/Kconfig
> +++ b/arch/loongarch/Kconfig
> @@ -143,7 +143,7 @@ config LOONGARCH
>         select HAVE_LIVEPATCH
>         select HAVE_MOD_ARCH_SPECIFIC
>         select HAVE_NMI
> -       select HAVE_OBJTOOL if AS_HAS_EXPLICIT_RELOCS
> +       select HAVE_OBJTOOL if AS_HAS_EXPLICIT_RELOCS && BROKEN
>         select HAVE_PCI
>         select HAVE_PERF_EVENTS
>         select HAVE_PERF_REGS
> 
> should be better?  It looks like everything depending on objtool is
> broken, not only ORC.

No longer needed, I've root-caused the issue:
https://lore.kernel.org/loongarch/20240604150741.30252-1-xry111@xry111.site/

-- 
Xi Ruoyao <xry111@xry111.site>
School of Aerospace Science and Technology, Xidian University