[2/2] sparc: Fix .udiv plt on libc
Commit Message
With the removal of divdi3 object from sparcv9-linux-gnu build, its
definition came from libgcc and its functions internall calls .udiv.
Since glibc also exports these symbols for compatibility reasons, it
will end up creating PLT calls internally in libc.so.
To avoid it, this patch uses the linker option --wrap to replace all
the internal libc.so .udiv calls to the wrapper __wrap_.udiv. Along
with strong alias in the udiv implementations, it makes linker do
local calls.
Checked on sparcv9-linux-gnu.
* sysdeps/sparc/sparc32/Makefile (libc.so-gnulib): New rule.
* sysdeps/sparc/sparc32/sparcv8/udiv.S (.udiv): Make a strong_alias
to __wrap_.udiv.
* sysdeps/sparc/sparc32/sparcv9/udiv.S (.udiv): Likewise.
* sysdeps/sparc/sparc32/udiv.S (.udiv): Likewise.
---
ChangeLog | 6 ++++++
sysdeps/sparc/sparc32/Makefile | 6 ++++++
sysdeps/sparc/sparc32/sparcv8/udiv.S | 1 +
sysdeps/sparc/sparc32/sparcv9/udiv.S | 1 +
sysdeps/sparc/sparc32/udiv.S | 1 +
5 files changed, 15 insertions(+)
Comments
On 03/16/2017 02:43 PM, Adhemerval Zanella wrote:
> To avoid it, this patch uses the linker option --wrap to replace all
> the internal libc.so .udiv calls to the wrapper __wrap_.udiv. Along
> with strong alias in the udiv implementations, it makes linker do
> local calls.
I think we already use a different mechanism for a similar purpose, via
sysdeps/generic/symbol-hacks.h. Maybe it's possible to reuse that?
Thanks,
Florian
On 16/03/2017 11:45, Florian Weimer wrote:
> On 03/16/2017 02:43 PM, Adhemerval Zanella wrote:
>> To avoid it, this patch uses the linker option --wrap to replace all
>> the internal libc.so .udiv calls to the wrapper __wrap_.udiv. Along
>> with strong alias in the udiv implementations, it makes linker do
>> local calls.
>
> I think we already use a different mechanism for a similar purpose, via sysdeps/generic/symbol-hacks.h. Maybe it's possible to reuse that?
>
> Thanks,
> Florian
The problem is now that divdi3 is not provided by glibc anymore we can't
use the same trick since __divdi3 and __moddi3 will be provided by libgcc.
Another way to actually redefine .udiv is to either still keep using divdi3
for sparc32 .
On 16/03/2017 13:59, Adhemerval Zanella wrote:
>
>
> On 16/03/2017 11:45, Florian Weimer wrote:
>> On 03/16/2017 02:43 PM, Adhemerval Zanella wrote:
>>> To avoid it, this patch uses the linker option --wrap to replace all
>>> the internal libc.so .udiv calls to the wrapper __wrap_.udiv. Along
>>> with strong alias in the udiv implementations, it makes linker do
>>> local calls.
>>
>> I think we already use a different mechanism for a similar purpose, via sysdeps/generic/symbol-hacks.h. Maybe it's possible to reuse that?
>>
>> Thanks,
>> Florian
>
> The problem is now that divdi3 is not provided by glibc anymore we can't
> use the same trick since __divdi3 and __moddi3 will be provided by libgcc.
> Another way to actually redefine .udiv is to either still keep using divdi3
> for sparc32 .
>
Any more impeding remark about this patch?
@@ -47,3 +47,9 @@ $(divrem:%=$(sysdep_dir)/sparc/sparc32/%.S): $(sysdep_dir)/sparc/sparc32/divrem.
mv -f $@-tmp $@
sysdep-realclean := $(sysdep-realclean) $(divrem:%=sysdeps/sparc/sparc32/%.S)
+
+# libgcc __divdi3 and __moddi3 use .udiv and since it is also exported by
+# libc.so linker will create PLTs for the symbol. To avoid it we strong alias
+# the exported libc one to __wrap_.udiv and use linker option --wrap to make any
+# call to .udiv to call the wrapper symbol.
+libc.so-gnulib += -Wl,--wrap=.udiv
@@ -13,3 +13,4 @@ ENTRY(.udiv)
udiv %o0, %o1, %o0
END(.udiv)
+strong_alias (.udiv, __wrap_.udiv)
@@ -15,3 +15,4 @@ ENTRY(.udiv)
udiv %o0, %o1, %o0
END(.udiv)
+strong_alias (.udiv, __wrap_.udiv)
@@ -344,3 +344,4 @@ LOC(got_result):
mov %o2, %o0
END(.udiv)
+strong_alias (.udiv, __wrap_.udiv)