clang-12 -fno-pic (since
https://github.com/llvm/llvm-project/commit/a084c0388e2a59b9556f2de0083333232da3f1d6)
can emit `call __stack_chk_fail@PLT` instead of `call __stack_chk_fail`
on x86. The two forms should have identical behaviors on x86-64 but the
former causes GNU as<2.37 to produce an unreferenced undefined symbol
_GLOBAL_OFFSET_TABLE_.
(On x86-32, there is an R_386_PC32 vs R_386_PLT32 difference but the
linker behavior is identical as far as Linux kernel is concerned.)
Simply ignore _GLOBAL_OFFSET_TABLE_ for now, like what
scripts/mod/modpost.c:ignore_undef_symbol does. This also fixes the
problem for gcc/clang -fpie and -fpic, which may emit `call foo@PLT` for
external function calls on x86.
Note: ld -z defs and dynamic loaders do not error for unreferenced
undefined symbols so the module loader is reading too much. If we ever
need to ignore more symbols, the code should be refactored to ignore
unreferenced symbols.
Reported-by: Marco Elver <[email protected]>
Link: https://github.com/ClangBuiltLinux/linux/issues/1250
Signed-off-by: Fangrui Song <[email protected]>
---
kernel/module.c | 10 ++++++++--
1 file changed, 8 insertions(+), 2 deletions(-)
diff --git a/kernel/module.c b/kernel/module.c
index 4bf30e4b3eaa..2e2deea99289 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -2395,8 +2395,14 @@ static int simplify_symbols(struct module *mod, const struct load_info *info)
break;
}
- /* Ok if weak. */
- if (!ksym && ELF_ST_BIND(sym[i].st_info) == STB_WEAK)
+ /* Ok if weak. Also allow _GLOBAL_OFFSET_TABLE_:
+ * GNU as before 2.37 produces an unreferenced _GLOBAL_OFFSET_TABLE_
+ * for call foo@PLT on x86-64. If the code ever needs to ignore
+ * more symbols, refactor the code to only warn if referenced by
+ * a relocation.
+ */
+ if (!ksym && (ELF_ST_BIND(sym[i].st_info) == STB_WEAK ||
+ !strcmp(name, "_GLOBAL_OFFSET_TABLE_")))
break;
ret = PTR_ERR(ksym) ?: -ENOENT;
--
2.30.0.284.gd98b1dd5eaa7-goog
+++ Fangrui Song [13/01/21 21:48 -0800]:
>clang-12 -fno-pic (since
>https://github.com/llvm/llvm-project/commit/a084c0388e2a59b9556f2de0083333232da3f1d6)
>can emit `call __stack_chk_fail@PLT` instead of `call __stack_chk_fail`
>on x86. The two forms should have identical behaviors on x86-64 but the
>former causes GNU as<2.37 to produce an unreferenced undefined symbol
>_GLOBAL_OFFSET_TABLE_.
>
>(On x86-32, there is an R_386_PC32 vs R_386_PLT32 difference but the
>linker behavior is identical as far as Linux kernel is concerned.)
>
>Simply ignore _GLOBAL_OFFSET_TABLE_ for now, like what
>scripts/mod/modpost.c:ignore_undef_symbol does. This also fixes the
>problem for gcc/clang -fpie and -fpic, which may emit `call foo@PLT` for
>external function calls on x86.
>
>Note: ld -z defs and dynamic loaders do not error for unreferenced
>undefined symbols so the module loader is reading too much. If we ever
>need to ignore more symbols, the code should be refactored to ignore
>unreferenced symbols.
>
>Reported-by: Marco Elver <[email protected]>
>Link: https://github.com/ClangBuiltLinux/linux/issues/1250
>Signed-off-by: Fangrui Song <[email protected]>
>---
> kernel/module.c | 10 ++++++++--
> 1 file changed, 8 insertions(+), 2 deletions(-)
>
>diff --git a/kernel/module.c b/kernel/module.c
>index 4bf30e4b3eaa..2e2deea99289 100644
>--- a/kernel/module.c
>+++ b/kernel/module.c
>@@ -2395,8 +2395,14 @@ static int simplify_symbols(struct module *mod, const struct load_info *info)
> break;
> }
>
>- /* Ok if weak. */
>- if (!ksym && ELF_ST_BIND(sym[i].st_info) == STB_WEAK)
>+ /* Ok if weak. Also allow _GLOBAL_OFFSET_TABLE_:
>+ * GNU as before 2.37 produces an unreferenced _GLOBAL_OFFSET_TABLE_
>+ * for call foo@PLT on x86-64. If the code ever needs to ignore
>+ * more symbols, refactor the code to only warn if referenced by
>+ * a relocation.
>+ */
>+ if (!ksym && (ELF_ST_BIND(sym[i].st_info) == STB_WEAK ||
>+ !strcmp(name, "_GLOBAL_OFFSET_TABLE_")))
> break;
Hi Fangrui,
Thanks for the patch. I am puzzled why we don't already mirror modpost
here, that particular line of code in modpost to ignore _GLOBAL_OFFSET_TABLE_
has been there long before my time. Let's properly mirror modpost
then, and create a similar helper function ignore_undef_symbol() (and
stick the _GLOBAL_OFFSET_TABLE_ check in there) to account for future
cases like this.
Thanks,
Jessica
On Thu, Jan 14, 2021 at 6:06 AM Jessica Yu <[email protected]> wrote:
>
> +++ Fangrui Song [13/01/21 21:48 -0800]:
> >clang-12 -fno-pic (since
> >https://github.com/llvm/llvm-project/commit/a084c0388e2a59b9556f2de0083333232da3f1d6)
> >can emit `call __stack_chk_fail@PLT` instead of `call __stack_chk_fail`
> >on x86. The two forms should have identical behaviors on x86-64 but the
> >former causes GNU as<2.37 to produce an unreferenced undefined symbol
> >_GLOBAL_OFFSET_TABLE_.
> >
> >(On x86-32, there is an R_386_PC32 vs R_386_PLT32 difference but the
> >linker behavior is identical as far as Linux kernel is concerned.)
> >
> >Simply ignore _GLOBAL_OFFSET_TABLE_ for now, like what
> >scripts/mod/modpost.c:ignore_undef_symbol does. This also fixes the
> >problem for gcc/clang -fpie and -fpic, which may emit `call foo@PLT` for
> >external function calls on x86.
> >
> >Note: ld -z defs and dynamic loaders do not error for unreferenced
> >undefined symbols so the module loader is reading too much. If we ever
> >need to ignore more symbols, the code should be refactored to ignore
> >unreferenced symbols.
> >
> >Reported-by: Marco Elver <[email protected]>
> >Link: https://github.com/ClangBuiltLinux/linux/issues/1250
> >Signed-off-by: Fangrui Song <[email protected]>
> >---
> > kernel/module.c | 10 ++++++++--
> > 1 file changed, 8 insertions(+), 2 deletions(-)
> >
> >diff --git a/kernel/module.c b/kernel/module.c
> >index 4bf30e4b3eaa..2e2deea99289 100644
> >--- a/kernel/module.c
> >+++ b/kernel/module.c
> >@@ -2395,8 +2395,14 @@ static int simplify_symbols(struct module *mod, const struct load_info *info)
> > break;
> > }
> >
> >- /* Ok if weak. */
> >- if (!ksym && ELF_ST_BIND(sym[i].st_info) == STB_WEAK)
> >+ /* Ok if weak. Also allow _GLOBAL_OFFSET_TABLE_:
> >+ * GNU as before 2.37 produces an unreferenced _GLOBAL_OFFSET_TABLE_
> >+ * for call foo@PLT on x86-64. If the code ever needs to ignore
> >+ * more symbols, refactor the code to only warn if referenced by
> >+ * a relocation.
> >+ */
> >+ if (!ksym && (ELF_ST_BIND(sym[i].st_info) == STB_WEAK ||
> >+ !strcmp(name, "_GLOBAL_OFFSET_TABLE_")))
> > break;
>
> Hi Fangrui,
>
> Thanks for the patch. I am puzzled why we don't already mirror modpost
> here, that particular line of code in modpost to ignore _GLOBAL_OFFSET_TABLE_
> has been there long before my time. Let's properly mirror modpost
> then, and create a similar helper function ignore_undef_symbol() (and
> stick the _GLOBAL_OFFSET_TABLE_ check in there) to account for future
> cases like this.
>
> Thanks,
>
> Jessica
Hi Jessica,
I guess __this_module in scripts/mod/modpost.c:ignore_undef_symbol is
not a problem.
For PPC64 _restgpr0_* and _savegpr0_*, I am not sure ignoring the
undefined functions in kernel/module.c is right.
(I know they can be produced by gcc -Os in some cases
(https://reviews.llvm.org/D79977), but I want to learn whether that is
a real issue before adding them.)
If we ever need to ignore more symbols, the code should be refactored
to not warn for unreferenced undefined symbols as my description says.
+++ Fāng-ruì Sòng [14/01/21 08:57 -0800]:
>On Thu, Jan 14, 2021 at 6:06 AM Jessica Yu <[email protected]> wrote:
>>
>> +++ Fangrui Song [13/01/21 21:48 -0800]:
>> >clang-12 -fno-pic (since
>> >https://github.com/llvm/llvm-project/commit/a084c0388e2a59b9556f2de0083333232da3f1d6)
>> >can emit `call __stack_chk_fail@PLT` instead of `call __stack_chk_fail`
>> >on x86. The two forms should have identical behaviors on x86-64 but the
>> >former causes GNU as<2.37 to produce an unreferenced undefined symbol
>> >_GLOBAL_OFFSET_TABLE_.
>> >
>> >(On x86-32, there is an R_386_PC32 vs R_386_PLT32 difference but the
>> >linker behavior is identical as far as Linux kernel is concerned.)
>> >
>> >Simply ignore _GLOBAL_OFFSET_TABLE_ for now, like what
>> >scripts/mod/modpost.c:ignore_undef_symbol does. This also fixes the
>> >problem for gcc/clang -fpie and -fpic, which may emit `call foo@PLT` for
>> >external function calls on x86.
>> >
>> >Note: ld -z defs and dynamic loaders do not error for unreferenced
>> >undefined symbols so the module loader is reading too much. If we ever
>> >need to ignore more symbols, the code should be refactored to ignore
>> >unreferenced symbols.
>> >
>> >Reported-by: Marco Elver <[email protected]>
>> >Link: https://github.com/ClangBuiltLinux/linux/issues/1250
>> >Signed-off-by: Fangrui Song <[email protected]>
>> >---
>> > kernel/module.c | 10 ++++++++--
>> > 1 file changed, 8 insertions(+), 2 deletions(-)
>> >
>> >diff --git a/kernel/module.c b/kernel/module.c
>> >index 4bf30e4b3eaa..2e2deea99289 100644
>> >--- a/kernel/module.c
>> >+++ b/kernel/module.c
>> >@@ -2395,8 +2395,14 @@ static int simplify_symbols(struct module *mod, const struct load_info *info)
>> > break;
>> > }
>> >
>> >- /* Ok if weak. */
>> >- if (!ksym && ELF_ST_BIND(sym[i].st_info) == STB_WEAK)
>> >+ /* Ok if weak. Also allow _GLOBAL_OFFSET_TABLE_:
>> >+ * GNU as before 2.37 produces an unreferenced _GLOBAL_OFFSET_TABLE_
>> >+ * for call foo@PLT on x86-64. If the code ever needs to ignore
>> >+ * more symbols, refactor the code to only warn if referenced by
>> >+ * a relocation.
>> >+ */
>> >+ if (!ksym && (ELF_ST_BIND(sym[i].st_info) == STB_WEAK ||
>> >+ !strcmp(name, "_GLOBAL_OFFSET_TABLE_")))
>> > break;
>>
>> Hi Fangrui,
>>
>> Thanks for the patch. I am puzzled why we don't already mirror modpost
>> here, that particular line of code in modpost to ignore _GLOBAL_OFFSET_TABLE_
>> has been there long before my time. Let's properly mirror modpost
>> then, and create a similar helper function ignore_undef_symbol() (and
>> stick the _GLOBAL_OFFSET_TABLE_ check in there) to account for future
>> cases like this.
>>
>> Thanks,
>>
>> Jessica
>
>Hi Jessica,
>
>I guess __this_module in scripts/mod/modpost.c:ignore_undef_symbol is
>not a problem.
>For PPC64 _restgpr0_* and _savegpr0_*, I am not sure ignoring the
>undefined functions in kernel/module.c is right.
>(I know they can be produced by gcc -Os in some cases
>(https://reviews.llvm.org/D79977), but I want to learn whether that is
>a real issue before adding them.)
>
>If we ever need to ignore more symbols, the code should be refactored
>to not warn for unreferenced undefined symbols as my description says.
Hi Fangrui,
Sorry for the confusion, I did not mean to exactly mirror
ignore_undef_symbol() in modpost. The conditions are different there
and not all of them would not apply to kernel/module.c. Like
__this_module, as you say, is not a problem as this will be resolved
once all the module *.o are linked in the final .ko. So when it
reaches the module loader it would no longer be UNDEF. I assume that
it is a similar situation for the PPC64 symbols.
What I meant that we could probably make this patch look nicer by
moving the hardcoded check for "_GLOBAL_OFFSET_TABLE_" to a helper
function ignore_undef_symbol(), similar to how modpost does it,
without adding any additional conditions for now. And yes, if we ever
need to extend this and refactor to not warn for unreferenced
undefined symbols, we should refactor to generalize this case, but for
the scope of this patch I think the suggested change is sufficient for
now.
Thank you,
Jessica
clang-12 -fno-pic (since
https://github.com/llvm/llvm-project/commit/a084c0388e2a59b9556f2de0083333232da3f1d6)
can emit `call __stack_chk_fail@PLT` instead of `call __stack_chk_fail`
on x86. The two forms should have identical behaviors on x86-64 but the
former causes GNU as<2.37 to produce an unreferenced undefined symbol
_GLOBAL_OFFSET_TABLE_.
(On x86-32, there is an R_386_PC32 vs R_386_PLT32 difference but the
linker behavior is identical as far as Linux kernel is concerned.)
Simply ignore _GLOBAL_OFFSET_TABLE_ for now, like what
scripts/mod/modpost.c:ignore_undef_symbol does. This also fixes the
problem for gcc/clang -fpie and -fpic, which may emit `call foo@PLT` for
external function calls on x86.
Note: ld -z defs and dynamic loaders do not error for unreferenced
undefined symbols so the module loader is reading too much. If we ever
need to ignore more symbols, the code should be refactored to ignore
unreferenced symbols.
Reported-by: Marco Elver <[email protected]>
Link: https://github.com/ClangBuiltLinux/linux/issues/1250
Signed-off-by: Fangrui Song <[email protected]>
---
kernel/module.c | 20 ++++++++++++++++++--
1 file changed, 18 insertions(+), 2 deletions(-)
---
Changes in v2:
* Fix Marco's email address
* Add a function ignore_undef_symbol similar to scripts/mod/modpost.c:ignore_undef_symbol
diff --git a/kernel/module.c b/kernel/module.c
index 4bf30e4b3eaa..278f5129bde2 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -2348,6 +2348,20 @@ static int verify_exported_symbols(struct module *mod)
return 0;
}
+static int ignore_undef_symbol(Elf_Half emachine, const char *name)
+{
+ /* On x86, PIC code and Clang non-PIC code may have call foo@PLT. GNU as
+ * before 2.37 produces an unreferenced _GLOBAL_OFFSET_TABLE_ on x86-64.
+ * i386 has a similar problem but may not deserve a fix.
+ *
+ * If we ever have to ignore many symbols, consider refactoring the code to
+ * only warn if referenced by a relocation.
+ */
+ if (emachine == EM_386 || emachine == EM_X86_64)
+ return !strcmp(name, "_GLOBAL_OFFSET_TABLE_");
+ return 0;
+}
+
/* Change all symbols so that st_value encodes the pointer directly. */
static int simplify_symbols(struct module *mod, const struct load_info *info)
{
@@ -2395,8 +2409,10 @@ static int simplify_symbols(struct module *mod, const struct load_info *info)
break;
}
- /* Ok if weak. */
- if (!ksym && ELF_ST_BIND(sym[i].st_info) == STB_WEAK)
+ /* Ok if weak or ignored. */
+ if (!ksym &&
+ (ELF_ST_BIND(sym[i].st_info) == STB_WEAK ||
+ ignore_undef_symbol(info->hdr->e_machine, name)))
break;
ret = PTR_ERR(ksym) ?: -ENOENT;
--
2.30.0.296.g2bfb1c46d8-goog
On Thu, Jan 14, 2021 at 1:54 PM 'Fangrui Song' via Clang Built Linux
<[email protected]> wrote:
>
> clang-12 -fno-pic (since
> https://github.com/llvm/llvm-project/commit/a084c0388e2a59b9556f2de0083333232da3f1d6)
> can emit `call __stack_chk_fail@PLT` instead of `call __stack_chk_fail`
> on x86. The two forms should have identical behaviors on x86-64 but the
> former causes GNU as<2.37 to produce an unreferenced undefined symbol
> _GLOBAL_OFFSET_TABLE_.
>
> (On x86-32, there is an R_386_PC32 vs R_386_PLT32 difference but the
> linker behavior is identical as far as Linux kernel is concerned.)
>
> Simply ignore _GLOBAL_OFFSET_TABLE_ for now, like what
> scripts/mod/modpost.c:ignore_undef_symbol does. This also fixes the
> problem for gcc/clang -fpie and -fpic, which may emit `call foo@PLT` for
> external function calls on x86.
>
> Note: ld -z defs and dynamic loaders do not error for unreferenced
> undefined symbols so the module loader is reading too much. If we ever
> need to ignore more symbols, the code should be refactored to ignore
> unreferenced symbols.
>
> Reported-by: Marco Elver <[email protected]>
> Link: https://github.com/ClangBuiltLinux/linux/issues/1250
> Signed-off-by: Fangrui Song <[email protected]>
Thanks for the patch.
Reviewed-by: Nick Desaulniers <[email protected]>
Jessica, would you mind adding when applying:
Cc: <[email protected]>
as I suspect we might want this fixed in stable tree's branches, too.
It might of interest to add:
Link: https://sourceware.org/bugzilla/show_bug.cgi?id=27178
too.
> ---
> kernel/module.c | 20 ++++++++++++++++++--
> 1 file changed, 18 insertions(+), 2 deletions(-)
> ---
> Changes in v2:
> * Fix Marco's email address
> * Add a function ignore_undef_symbol similar to scripts/mod/modpost.c:ignore_undef_symbol
>
> diff --git a/kernel/module.c b/kernel/module.c
> index 4bf30e4b3eaa..278f5129bde2 100644
> --- a/kernel/module.c
> +++ b/kernel/module.c
> @@ -2348,6 +2348,20 @@ static int verify_exported_symbols(struct module *mod)
> return 0;
> }
>
> +static int ignore_undef_symbol(Elf_Half emachine, const char *name)
> +{
> + /* On x86, PIC code and Clang non-PIC code may have call foo@PLT. GNU as
> + * before 2.37 produces an unreferenced _GLOBAL_OFFSET_TABLE_ on x86-64.
> + * i386 has a similar problem but may not deserve a fix.
> + *
> + * If we ever have to ignore many symbols, consider refactoring the code to
> + * only warn if referenced by a relocation.
> + */
> + if (emachine == EM_386 || emachine == EM_X86_64)
> + return !strcmp(name, "_GLOBAL_OFFSET_TABLE_");
> + return 0;
> +}
> +
> /* Change all symbols so that st_value encodes the pointer directly. */
> static int simplify_symbols(struct module *mod, const struct load_info *info)
> {
> @@ -2395,8 +2409,10 @@ static int simplify_symbols(struct module *mod, const struct load_info *info)
> break;
> }
>
> - /* Ok if weak. */
> - if (!ksym && ELF_ST_BIND(sym[i].st_info) == STB_WEAK)
> + /* Ok if weak or ignored. */
> + if (!ksym &&
> + (ELF_ST_BIND(sym[i].st_info) == STB_WEAK ||
> + ignore_undef_symbol(info->hdr->e_machine, name)))
> break;
>
> ret = PTR_ERR(ksym) ?: -ENOENT;
> --
--
Thanks,
~Nick Desaulniers
On Thu, 14 Jan 2021 at 22:54, Fangrui Song <[email protected]> wrote:
> clang-12 -fno-pic (since
> https://github.com/llvm/llvm-project/commit/a084c0388e2a59b9556f2de0083333232da3f1d6)
> can emit `call __stack_chk_fail@PLT` instead of `call __stack_chk_fail`
> on x86. The two forms should have identical behaviors on x86-64 but the
> former causes GNU as<2.37 to produce an unreferenced undefined symbol
> _GLOBAL_OFFSET_TABLE_.
>
> (On x86-32, there is an R_386_PC32 vs R_386_PLT32 difference but the
> linker behavior is identical as far as Linux kernel is concerned.)
>
> Simply ignore _GLOBAL_OFFSET_TABLE_ for now, like what
> scripts/mod/modpost.c:ignore_undef_symbol does. This also fixes the
> problem for gcc/clang -fpie and -fpic, which may emit `call foo@PLT` for
> external function calls on x86.
>
> Note: ld -z defs and dynamic loaders do not error for unreferenced
> undefined symbols so the module loader is reading too much. If we ever
> need to ignore more symbols, the code should be refactored to ignore
> unreferenced symbols.
>
> Reported-by: Marco Elver <[email protected]>
> Link: https://github.com/ClangBuiltLinux/linux/issues/1250
> Signed-off-by: Fangrui Song <[email protected]>
Tested-by: Marco Elver <[email protected]>
Thank you for the patch!
> ---
> kernel/module.c | 20 ++++++++++++++++++--
> 1 file changed, 18 insertions(+), 2 deletions(-)
> ---
> Changes in v2:
> * Fix Marco's email address
> * Add a function ignore_undef_symbol similar to scripts/mod/modpost.c:ignore_undef_symbol
>
> diff --git a/kernel/module.c b/kernel/module.c
> index 4bf30e4b3eaa..278f5129bde2 100644
> --- a/kernel/module.c
> +++ b/kernel/module.c
> @@ -2348,6 +2348,20 @@ static int verify_exported_symbols(struct module *mod)
> return 0;
> }
>
> +static int ignore_undef_symbol(Elf_Half emachine, const char *name)
Why not 'bool' return-type?
> +{
> + /* On x86, PIC code and Clang non-PIC code may have call foo@PLT. GNU as
Not sure if checkpatch.pl warns about this, but this multi-line
comment does not follow the normal kernel-style (see elsewhere in
file):
/*
* ...
*/
> + * before 2.37 produces an unreferenced _GLOBAL_OFFSET_TABLE_ on x86-64.
> + * i386 has a similar problem but may not deserve a fix.
> + *
> + * If we ever have to ignore many symbols, consider refactoring the code to
> + * only warn if referenced by a relocation.
> + */
> + if (emachine == EM_386 || emachine == EM_X86_64)
> + return !strcmp(name, "_GLOBAL_OFFSET_TABLE_");
> + return 0;
> +}
> +
> /* Change all symbols so that st_value encodes the pointer directly. */
> static int simplify_symbols(struct module *mod, const struct load_info *info)
> {
> @@ -2395,8 +2409,10 @@ static int simplify_symbols(struct module *mod, const struct load_info *info)
> break;
> }
>
> - /* Ok if weak. */
> - if (!ksym && ELF_ST_BIND(sym[i].st_info) == STB_WEAK)
> + /* Ok if weak or ignored. */
> + if (!ksym &&
> + (ELF_ST_BIND(sym[i].st_info) == STB_WEAK ||
> + ignore_undef_symbol(info->hdr->e_machine, name)))
> break;
>
> ret = PTR_ERR(ksym) ?: -ENOENT;
> --
> 2.30.0.296.g2bfb1c46d8-goog
>
+++ Nick Desaulniers [14/01/21 14:01 -0800]:
>On Thu, Jan 14, 2021 at 1:54 PM 'Fangrui Song' via Clang Built Linux
><[email protected]> wrote:
>>
>> clang-12 -fno-pic (since
>> https://github.com/llvm/llvm-project/commit/a084c0388e2a59b9556f2de0083333232da3f1d6)
>> can emit `call __stack_chk_fail@PLT` instead of `call __stack_chk_fail`
>> on x86. The two forms should have identical behaviors on x86-64 but the
>> former causes GNU as<2.37 to produce an unreferenced undefined symbol
>> _GLOBAL_OFFSET_TABLE_.
>>
>> (On x86-32, there is an R_386_PC32 vs R_386_PLT32 difference but the
>> linker behavior is identical as far as Linux kernel is concerned.)
>>
>> Simply ignore _GLOBAL_OFFSET_TABLE_ for now, like what
>> scripts/mod/modpost.c:ignore_undef_symbol does. This also fixes the
>> problem for gcc/clang -fpie and -fpic, which may emit `call foo@PLT` for
>> external function calls on x86.
>>
>> Note: ld -z defs and dynamic loaders do not error for unreferenced
>> undefined symbols so the module loader is reading too much. If we ever
>> need to ignore more symbols, the code should be refactored to ignore
>> unreferenced symbols.
>>
>> Reported-by: Marco Elver <[email protected]>
>> Link: https://github.com/ClangBuiltLinux/linux/issues/1250
>> Signed-off-by: Fangrui Song <[email protected]>
>
>Thanks for the patch.
>
>Reviewed-by: Nick Desaulniers <[email protected]>
>
>Jessica, would you mind adding when applying:
>
>Cc: <[email protected]>
>
>as I suspect we might want this fixed in stable tree's branches, too.
>It might of interest to add:
>
>Link: https://sourceware.org/bugzilla/show_bug.cgi?id=27178
>
>too.
Sure, will do!
Thanks,
Jessica
+++ Marco Elver [15/01/21 08:03 +0100]:
>On Thu, 14 Jan 2021 at 22:54, Fangrui Song <[email protected]> wrote:
>> clang-12 -fno-pic (since
>> https://github.com/llvm/llvm-project/commit/a084c0388e2a59b9556f2de0083333232da3f1d6)
>> can emit `call __stack_chk_fail@PLT` instead of `call __stack_chk_fail`
>> on x86. The two forms should have identical behaviors on x86-64 but the
>> former causes GNU as<2.37 to produce an unreferenced undefined symbol
>> _GLOBAL_OFFSET_TABLE_.
>>
>> (On x86-32, there is an R_386_PC32 vs R_386_PLT32 difference but the
>> linker behavior is identical as far as Linux kernel is concerned.)
>>
>> Simply ignore _GLOBAL_OFFSET_TABLE_ for now, like what
>> scripts/mod/modpost.c:ignore_undef_symbol does. This also fixes the
>> problem for gcc/clang -fpie and -fpic, which may emit `call foo@PLT` for
>> external function calls on x86.
>>
>> Note: ld -z defs and dynamic loaders do not error for unreferenced
>> undefined symbols so the module loader is reading too much. If we ever
>> need to ignore more symbols, the code should be refactored to ignore
>> unreferenced symbols.
>>
>> Reported-by: Marco Elver <[email protected]>
>> Link: https://github.com/ClangBuiltLinux/linux/issues/1250
>> Signed-off-by: Fangrui Song <[email protected]>
>
>Tested-by: Marco Elver <[email protected]>
>
>Thank you for the patch!
>
>> ---
>> kernel/module.c | 20 ++++++++++++++++++--
>> 1 file changed, 18 insertions(+), 2 deletions(-)
>> ---
>> Changes in v2:
>> * Fix Marco's email address
>> * Add a function ignore_undef_symbol similar to scripts/mod/modpost.c:ignore_undef_symbol
>>
>> diff --git a/kernel/module.c b/kernel/module.c
>> index 4bf30e4b3eaa..278f5129bde2 100644
>> --- a/kernel/module.c
>> +++ b/kernel/module.c
>> @@ -2348,6 +2348,20 @@ static int verify_exported_symbols(struct module *mod)
>> return 0;
>> }
>>
>> +static int ignore_undef_symbol(Elf_Half emachine, const char *name)
>
>Why not 'bool' return-type?
>
>> +{
>> + /* On x86, PIC code and Clang non-PIC code may have call foo@PLT. GNU as
>
>Not sure if checkpatch.pl warns about this, but this multi-line
>comment does not follow the normal kernel-style (see elsewhere in
>file):
>
>/*
> * ...
> */
+1 to Marco's comments. Otherwise, patch looks good to me.
Thanks Fangrui!
Jessica
On Thu, Jan 14, 2021 at 11:04 PM Marco Elver <[email protected]> wrote:
>
> On Thu, 14 Jan 2021 at 22:54, Fangrui Song <[email protected]> wrote:
> > clang-12 -fno-pic (since
> > https://github.com/llvm/llvm-project/commit/a084c0388e2a59b9556f2de0083333232da3f1d6)
> > can emit `call __stack_chk_fail@PLT` instead of `call __stack_chk_fail`
> > on x86. The two forms should have identical behaviors on x86-64 but the
> > former causes GNU as<2.37 to produce an unreferenced undefined symbol
> > _GLOBAL_OFFSET_TABLE_.
> >
> > (On x86-32, there is an R_386_PC32 vs R_386_PLT32 difference but the
> > linker behavior is identical as far as Linux kernel is concerned.)
> >
> > Simply ignore _GLOBAL_OFFSET_TABLE_ for now, like what
> > scripts/mod/modpost.c:ignore_undef_symbol does. This also fixes the
> > problem for gcc/clang -fpie and -fpic, which may emit `call foo@PLT` for
> > external function calls on x86.
> >
> > Note: ld -z defs and dynamic loaders do not error for unreferenced
> > undefined symbols so the module loader is reading too much. If we ever
> > need to ignore more symbols, the code should be refactored to ignore
> > unreferenced symbols.
> >
> > Reported-by: Marco Elver <[email protected]>
> > Link: https://github.com/ClangBuiltLinux/linux/issues/1250
> > Signed-off-by: Fangrui Song <[email protected]>
>
> Tested-by: Marco Elver <[email protected]>
>
> Thank you for the patch!
>
> > ---
> > kernel/module.c | 20 ++++++++++++++++++--
> > 1 file changed, 18 insertions(+), 2 deletions(-)
> > ---
> > Changes in v2:
> > * Fix Marco's email address
> > * Add a function ignore_undef_symbol similar to scripts/mod/modpost.c:ignore_undef_symbol
> >
> > diff --git a/kernel/module.c b/kernel/module.c
> > index 4bf30e4b3eaa..278f5129bde2 100644
> > --- a/kernel/module.c
> > +++ b/kernel/module.c
> > @@ -2348,6 +2348,20 @@ static int verify_exported_symbols(struct module *mod)
> > return 0;
> > }
> >
> > +static int ignore_undef_symbol(Elf_Half emachine, const char *name)
>
> Why not 'bool' return-type?
Will use bool and false in v3.
> > +{
> > + /* On x86, PIC code and Clang non-PIC code may have call foo@PLT. GNU as
>
> Not sure if checkpatch.pl warns about this, but this multi-line
> comment does not follow the normal kernel-style (see elsewhere in
> file):
It doesn't warn about this. (The commit description warning cannot be
fixed, even if I place the closing paren on the next line.)
% ./scripts/checkpatch.pl
v2-0001-module-Ignore-_GLOBAL_OFFSET_TABLE_-when-warning-.patch
WARNING: Possible unwrapped commit description (prefer a maximum 75
chars per line)
#8:
https://github.com/llvm/llvm-project/commit/a084c0388e2a59b9556f2de0083333232da3f1d6)
total: 0 errors, 1 warnings, 32 lines checked
NOTE: For some of the reported defects, checkpatch may be able to
mechanically convert to the typical style using --fix or --fix-inplace.
v2-0001-module-Ignore-_GLOBAL_OFFSET_TABLE_-when-warning-.patch has
style problems, please review.
NOTE: If any of the errors are false positives, please report
them to the maintainer, see CHECKPATCH in MAINTAINERS.
> /*
> * ...
> */
>
> > + * before 2.37 produces an unreferenced _GLOBAL_OFFSET_TABLE_ on x86-64.
> > + * i386 has a similar problem but may not deserve a fix.
> > + *
> > + * If we ever have to ignore many symbols, consider refactoring the code to
> > + * only warn if referenced by a relocation.
> > + */
> > + if (emachine == EM_386 || emachine == EM_X86_64)
> > + return !strcmp(name, "_GLOBAL_OFFSET_TABLE_");
> > + return 0;
> > +}
> > +
> > /* Change all symbols so that st_value encodes the pointer directly. */
> > static int simplify_symbols(struct module *mod, const struct load_info *info)
> > {
> > @@ -2395,8 +2409,10 @@ static int simplify_symbols(struct module *mod, const struct load_info *info)
> > break;
> > }
> >
> > - /* Ok if weak. */
> > - if (!ksym && ELF_ST_BIND(sym[i].st_info) == STB_WEAK)
> > + /* Ok if weak or ignored. */
> > + if (!ksym &&
> > + (ELF_ST_BIND(sym[i].st_info) == STB_WEAK ||
> > + ignore_undef_symbol(info->hdr->e_machine, name)))
> > break;
> >
> > ret = PTR_ERR(ksym) ?: -ENOENT;
> > --
> > 2.30.0.296.g2bfb1c46d8-goog
> >
--
宋方睿
clang-12 -fno-pic (since
https://github.com/llvm/llvm-project/commit/a084c0388e2a59b9556f2de0083333232da3f1d6)
can emit `call __stack_chk_fail@PLT` instead of `call __stack_chk_fail`
on x86. The two forms should have identical behaviors on x86-64 but the
former causes GNU as<2.37 to produce an unreferenced undefined symbol
_GLOBAL_OFFSET_TABLE_.
(On x86-32, there is an R_386_PC32 vs R_386_PLT32 difference but the
linker behavior is identical as far as Linux kernel is concerned.)
Simply ignore _GLOBAL_OFFSET_TABLE_ for now, like what
scripts/mod/modpost.c:ignore_undef_symbol does. This also fixes the
problem for gcc/clang -fpie and -fpic, which may emit `call foo@PLT` for
external function calls on x86.
Note: ld -z defs and dynamic loaders do not error for unreferenced
undefined symbols so the module loader is reading too much. If we ever
need to ignore more symbols, the code should be refactored to ignore
unreferenced symbols.
Reported-by: Marco Elver <[email protected]>
Link: https://github.com/ClangBuiltLinux/linux/issues/1250
Signed-off-by: Fangrui Song <[email protected]>
Reviewed-by: Nick Desaulniers <[email protected]>
Tested-by: Marco Elver <[email protected]>
Cc: <[email protected]>
---
Changes in v2:
* Fix Marco's email address
* Add a function ignore_undef_symbol similar to scripts/mod/modpost.c:ignore_undef_symbol
---
Changes in v3:
* Fix the style of a multi-line comment.
* Use static bool ignore_undef_symbol.
---
kernel/module.c | 21 +++++++++++++++++++--
1 file changed, 19 insertions(+), 2 deletions(-)
diff --git a/kernel/module.c b/kernel/module.c
index 4bf30e4b3eaa..805c49d1b86d 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -2348,6 +2348,21 @@ static int verify_exported_symbols(struct module *mod)
return 0;
}
+static bool ignore_undef_symbol(Elf_Half emachine, const char *name)
+{
+ /*
+ * On x86, PIC code and Clang non-PIC code may have call foo@PLT. GNU as
+ * before 2.37 produces an unreferenced _GLOBAL_OFFSET_TABLE_ on x86-64.
+ * i386 has a similar problem but may not deserve a fix.
+ *
+ * If we ever have to ignore many symbols, consider refactoring the code to
+ * only warn if referenced by a relocation.
+ */
+ if (emachine == EM_386 || emachine == EM_X86_64)
+ return !strcmp(name, "_GLOBAL_OFFSET_TABLE_");
+ return false;
+}
+
/* Change all symbols so that st_value encodes the pointer directly. */
static int simplify_symbols(struct module *mod, const struct load_info *info)
{
@@ -2395,8 +2410,10 @@ static int simplify_symbols(struct module *mod, const struct load_info *info)
break;
}
- /* Ok if weak. */
- if (!ksym && ELF_ST_BIND(sym[i].st_info) == STB_WEAK)
+ /* Ok if weak or ignored. */
+ if (!ksym &&
+ (ELF_ST_BIND(sym[i].st_info) == STB_WEAK ||
+ ignore_undef_symbol(info->hdr->e_machine, name)))
break;
ret = PTR_ERR(ksym) ?: -ENOENT;
--
2.30.0.296.g2bfb1c46d8-goog
On Fri, Jan 15, 2021 at 11:52:22AM -0800, 'Fangrui Song' via Clang Built Linux wrote:
> clang-12 -fno-pic (since
> https://github.com/llvm/llvm-project/commit/a084c0388e2a59b9556f2de0083333232da3f1d6)
> can emit `call __stack_chk_fail@PLT` instead of `call __stack_chk_fail`
> on x86. The two forms should have identical behaviors on x86-64 but the
> former causes GNU as<2.37 to produce an unreferenced undefined symbol
> _GLOBAL_OFFSET_TABLE_.
>
> (On x86-32, there is an R_386_PC32 vs R_386_PLT32 difference but the
> linker behavior is identical as far as Linux kernel is concerned.)
>
> Simply ignore _GLOBAL_OFFSET_TABLE_ for now, like what
> scripts/mod/modpost.c:ignore_undef_symbol does. This also fixes the
> problem for gcc/clang -fpie and -fpic, which may emit `call foo@PLT` for
> external function calls on x86.
>
> Note: ld -z defs and dynamic loaders do not error for unreferenced
> undefined symbols so the module loader is reading too much. If we ever
> need to ignore more symbols, the code should be refactored to ignore
> unreferenced symbols.
>
> Reported-by: Marco Elver <[email protected]>
> Link: https://github.com/ClangBuiltLinux/linux/issues/1250
> Signed-off-by: Fangrui Song <[email protected]>
> Reviewed-by: Nick Desaulniers <[email protected]>
> Tested-by: Marco Elver <[email protected]>
> Cc: <[email protected]>
Reviewed-by: Nathan Chancellor <[email protected]>
>
> ---
> Changes in v2:
> * Fix Marco's email address
> * Add a function ignore_undef_symbol similar to scripts/mod/modpost.c:ignore_undef_symbol
> ---
> Changes in v3:
> * Fix the style of a multi-line comment.
> * Use static bool ignore_undef_symbol.
> ---
> kernel/module.c | 21 +++++++++++++++++++--
> 1 file changed, 19 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/module.c b/kernel/module.c
> index 4bf30e4b3eaa..805c49d1b86d 100644
> --- a/kernel/module.c
> +++ b/kernel/module.c
> @@ -2348,6 +2348,21 @@ static int verify_exported_symbols(struct module *mod)
> return 0;
> }
>
> +static bool ignore_undef_symbol(Elf_Half emachine, const char *name)
> +{
> + /*
> + * On x86, PIC code and Clang non-PIC code may have call foo@PLT. GNU as
> + * before 2.37 produces an unreferenced _GLOBAL_OFFSET_TABLE_ on x86-64.
> + * i386 has a similar problem but may not deserve a fix.
> + *
> + * If we ever have to ignore many symbols, consider refactoring the code to
> + * only warn if referenced by a relocation.
> + */
> + if (emachine == EM_386 || emachine == EM_X86_64)
> + return !strcmp(name, "_GLOBAL_OFFSET_TABLE_");
> + return false;
> +}
> +
> /* Change all symbols so that st_value encodes the pointer directly. */
> static int simplify_symbols(struct module *mod, const struct load_info *info)
> {
> @@ -2395,8 +2410,10 @@ static int simplify_symbols(struct module *mod, const struct load_info *info)
> break;
> }
>
> - /* Ok if weak. */
> - if (!ksym && ELF_ST_BIND(sym[i].st_info) == STB_WEAK)
> + /* Ok if weak or ignored. */
> + if (!ksym &&
> + (ELF_ST_BIND(sym[i].st_info) == STB_WEAK ||
> + ignore_undef_symbol(info->hdr->e_machine, name)))
> break;
>
> ret = PTR_ERR(ksym) ?: -ENOENT;
> --
> 2.30.0.296.g2bfb1c46d8-goog
>
> --
> You received this message because you are subscribed to the Google Groups "Clang Built Linux" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
> To view this discussion on the web visit https://groups.google.com/d/msgid/clang-built-linux/20210115195222.3453262-1-maskray%40google.com.
+++ Fangrui Song [15/01/21 11:52 -0800]:
>clang-12 -fno-pic (since
>https://github.com/llvm/llvm-project/commit/a084c0388e2a59b9556f2de0083333232da3f1d6)
>can emit `call __stack_chk_fail@PLT` instead of `call __stack_chk_fail`
>on x86. The two forms should have identical behaviors on x86-64 but the
>former causes GNU as<2.37 to produce an unreferenced undefined symbol
>_GLOBAL_OFFSET_TABLE_.
>
>(On x86-32, there is an R_386_PC32 vs R_386_PLT32 difference but the
>linker behavior is identical as far as Linux kernel is concerned.)
>
>Simply ignore _GLOBAL_OFFSET_TABLE_ for now, like what
>scripts/mod/modpost.c:ignore_undef_symbol does. This also fixes the
>problem for gcc/clang -fpie and -fpic, which may emit `call foo@PLT` for
>external function calls on x86.
>
>Note: ld -z defs and dynamic loaders do not error for unreferenced
>undefined symbols so the module loader is reading too much. If we ever
>need to ignore more symbols, the code should be refactored to ignore
>unreferenced symbols.
>
>Reported-by: Marco Elver <[email protected]>
>Link: https://github.com/ClangBuiltLinux/linux/issues/1250
>Signed-off-by: Fangrui Song <[email protected]>
>Reviewed-by: Nick Desaulniers <[email protected]>
>Tested-by: Marco Elver <[email protected]>
>Cc: <[email protected]>
>
>---
>Changes in v2:
>* Fix Marco's email address
>* Add a function ignore_undef_symbol similar to scripts/mod/modpost.c:ignore_undef_symbol
>---
>Changes in v3:
>* Fix the style of a multi-line comment.
>* Use static bool ignore_undef_symbol.
Patch has been queued up on modules-next:
https://git.kernel.org/pub/scm/linux/kernel/git/jeyu/linux.git/commit/?h=modules-next&id=ebfac7b778fac8b0e8e92ec91d0b055f046b4604
Thanks!
Jessica