[PATCH] x86/purgatory: Compile purgatory with -D__DISABLE_EXPORTS

[PATCH] x86/purgatory: Compile purgatory with -D__DISABLE_EXPORTS

[Arnd Bergmann]

From: Arnd Bergmann <arnd@arndb.de>

A previous patch fixed compile-time issues by setting __NO_FORTIFY
for all of the purgatory code, but this is insufficient and still
leaves build failures that appeared recently in linux-next:

ld.lld: error: undefined symbol: __memset
>>> referenced by string.c
>>>               arch/x86/purgatory/purgatory.ro:(__UNIQUE_ID_addressable___memset_2)

ld.lld: error: undefined symbol: __memmove
>>> referenced by string.c
>>>               arch/x86/purgatory/purgatory.ro:(__UNIQUE_ID_addressable___memmove_3)

Set __DISABLE_EXPORTS as well, to avoid pulling in a reference to the
wrong string functions.

Fixes: 65cbe410d312 ("x86/purgatory: Compile purgatory with -D__NO_FORTIFY")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
 arch/x86/purgatory/Makefile | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/arch/x86/purgatory/Makefile b/arch/x86/purgatory/Makefile
index cde8463b4d84..df7dbb5b3387 100644
--- a/arch/x86/purgatory/Makefile
+++ b/arch/x86/purgatory/Makefile
@@ -11,8 +11,6 @@ $(obj)/string.o: $(srctree)/arch/x86/boot/compressed/string.c FORCE
 $(obj)/sha256.o: $(srctree)/lib/crypto/sha256.c FORCE
 	$(call if_changed_rule,cc_o_c)
 
-CFLAGS_sha256.o := -D__DISABLE_EXPORTS
-
 # When profile-guided optimization is enabled, llvm emits two different
 # overlapping text sections, which is not supported by kexec. Remove profile
 # optimization flags.
@@ -37,7 +35,7 @@ PURGATORY_CFLAGS := -mcmodel=small -ffreestanding -fno-zero-initialized-in-bss -
 PURGATORY_CFLAGS += -fpic -fvisibility=hidden
 PURGATORY_CFLAGS += $(DISABLE_KSTACK_ERASE) -DDISABLE_BRANCH_PROFILING
 PURGATORY_CFLAGS += -fno-stack-protector
-PURGATORY_CFLAGS += -D__NO_FORTIFY
+PURGATORY_CFLAGS += -D__NO_FORTIFY -D__DISABLE_EXPORTS
 
 # Default KBUILD_CFLAGS can have -pg option set when FTRACE is enabled. That
 # in turn leaves some undefined symbols like __fentry__ in purgatory and not
-- 
2.53.0

Re: [PATCH] x86/purgatory: Compile purgatory with -D__DISABLE_EXPORTS

[Eric Biggers]

On Fri, Sep 25, 2026 at 03:01:49PM +0200, Arnd Bergmann wrote:
> From: Arnd Bergmann <arnd@arndb.de>
> 
> A previous patch fixed compile-time issues by setting __NO_FORTIFY
> for all of the purgatory code, but this is insufficient and still
> leaves build failures that appeared recently in linux-next:
> 
> ld.lld: error: undefined symbol: __memset
> >>> referenced by string.c
> >>>               arch/x86/purgatory/purgatory.ro:(__UNIQUE_ID_addressable___memset_2)
> 
> ld.lld: error: undefined symbol: __memmove
> >>> referenced by string.c
> >>>               arch/x86/purgatory/purgatory.ro:(__UNIQUE_ID_addressable___memmove_3)
> 
> Set __DISABLE_EXPORTS as well, to avoid pulling in a reference to the
> wrong string functions.
> 
> Fixes: 65cbe410d312 ("x86/purgatory: Compile purgatory with -D__NO_FORTIFY")
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---
>  arch/x86/purgatory/Makefile | 4 +---
>  1 file changed, 1 insertion(+), 3 deletions(-)
> 

I dropped the patch this fixes already, as I mentioned at
https://lore.kernel.org/linux-crypto/20260924174911.GA1978@sol/ .  It's
gone from linux-next now.  I think Thomas Huth is planning to send a
fixed version, together with the two patches that depend on this.

But compiling all of purgatory (and not just sha256.o as is the status
quo) with -D__NO_FORTIFY -D__DISABLE_EXPORTS sounds good to me.

Alternatively it could be just sha256.o and purgatory.o specifically.
But it seems it might as well just be all files.

- Eric

Re: [PATCH] x86/purgatory: Compile purgatory with -D__DISABLE_EXPORTS

[Thomas Huth]

On 25/09/2026 21.16, Eric Biggers wrote:
> On Fri, Sep 25, 2026 at 03:01:49PM +0200, Arnd Bergmann wrote:
>> From: Arnd Bergmann <arnd@arndb.de>
>>
>> A previous patch fixed compile-time issues by setting __NO_FORTIFY
>> for all of the purgatory code, but this is insufficient and still
>> leaves build failures that appeared recently in linux-next:
>>
>> ld.lld: error: undefined symbol: __memset
>>>>> referenced by string.c
>>>>>                arch/x86/purgatory/purgatory.ro:(__UNIQUE_ID_addressable___memset_2)
>>
>> ld.lld: error: undefined symbol: __memmove
>>>>> referenced by string.c
>>>>>                arch/x86/purgatory/purgatory.ro:(__UNIQUE_ID_addressable___memmove_3)
>>
>> Set __DISABLE_EXPORTS as well, to avoid pulling in a reference to the
>> wrong string functions.
>>
>> Fixes: 65cbe410d312 ("x86/purgatory: Compile purgatory with -D__NO_FORTIFY")
>> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
>> ---
>>   arch/x86/purgatory/Makefile | 4 +---
>>   1 file changed, 1 insertion(+), 3 deletions(-)
>>
> 
> I dropped the patch this fixes already, as I mentioned at
> https://lore.kernel.org/linux-crypto/20260924174911.GA1978@sol/ .  It's
> gone from linux-next now.  I think Thomas Huth is planning to send a
> fixed version, together with the two patches that depend on this.
> 
> But compiling all of purgatory (and not just sha256.o as is the status
> quo) with -D__NO_FORTIFY -D__DISABLE_EXPORTS sounds good to me.
Yes, I think that's the way to go, and I've now also verified that the 
change is working. I've now send a new version of the patch here:

  https://lore.kernel.org/lkml/20260928083055.104266-2-thuth@redhat.com/

  Thomas