| Message ID | alpine.DEB.2.21.1907300047220.1468@digraph.polyomino.org.uk |
|---|---|
| State | New |
| Headers | show |
| Series | Restore r31 setting in powerpc32 swapcontext | expand |
* Joseph Myers: > diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S b/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S > index efebb10bba..6fa1ab7d6e 100644 > --- a/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S > +++ b/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S > @@ -276,6 +276,9 @@ ENTRY(__CONTEXT_FUNC_NAME) > cmpwi r3,0 > bne 3f /* L(error_exit) */ > > + lwz r4,_FRAME_PARM_SAVE2(r1) > + lwz r31,_UC_REGS_PTR(r4) > + > #ifdef __CONTEXT_ENABLE_FPRS > # ifdef __CONTEXT_ENABLE_VRS Patch looks good to me. Carlos will have to ack it as the RM, though. Thanks, Florian
On 29/07/2019 21:48, Joseph Myers wrote: > Commit ffe8a9a8318e1db225b22da8bc067408494bac5c, "powerpc: Remove > rt_sigreturn usage on context function", removed from powerpc32 > swapcontext a setting of r31 that is relied upon in subsequent code. > I'm not sure why this didn't produce test failures in Adhemerval's > 32-bit testing; in my (soft-float) testing in preparation for 2.30 > release, I see several context-related failures Thanks for catching it, I am trying to understand why I didn't see it on the environment I am using (gcc 8.1 on a POWER7). I just tested for hard-float, I will try a soft-float to see if it is related. I just tested with 1ff1373b3302 (which is the latest commit I checked on the machine) and the tests do pass... > > FAIL: stdlib/tst-makecontext2 > FAIL: stdlib/tst-makecontext3 > FAIL: stdlib/tst-setcontext > FAIL: stdlib/tst-setcontext2 > FAIL: stdlib/tst-setcontext4 > FAIL: stdlib/tst-setcontext7 > FAIL: stdlib/tst-setcontext9 > FAIL: stdlib/tst-swapcontext1 > > that did not appear in 2.29 testing. This patch restores the removed > register setting in question, and thus fixes those failures. > > Tested for powerpc (soft-float). > > 2019-07-30 Joseph Myers <joseph@codesourcery.com> > > * sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S > (__CONTEXT_FUNC_NAME): Restore setting of r31. > > diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S b/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S > index efebb10bba..6fa1ab7d6e 100644 > --- a/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S > +++ b/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S > @@ -276,6 +276,9 @@ ENTRY(__CONTEXT_FUNC_NAME) > cmpwi r3,0 > bne 3f /* L(error_exit) */ > > + lwz r4,_FRAME_PARM_SAVE2(r1) > + lwz r31,_UC_REGS_PTR(r4) > + > #ifdef __CONTEXT_ENABLE_FPRS > # ifdef __CONTEXT_ENABLE_VRS > >
On 30/07/2019 09:17, Adhemerval Zanella wrote: > > > On 29/07/2019 21:48, Joseph Myers wrote: >> Commit ffe8a9a8318e1db225b22da8bc067408494bac5c, "powerpc: Remove >> rt_sigreturn usage on context function", removed from powerpc32 >> swapcontext a setting of r31 that is relied upon in subsequent code. >> I'm not sure why this didn't produce test failures in Adhemerval's >> 32-bit testing; in my (soft-float) testing in preparation for 2.30 >> release, I see several context-related failures > > Thanks for catching it, I am trying to understand why I didn't see it on the > environment I am using (gcc 8.1 on a POWER7). I just tested for hard-float, > I will try a soft-float to see if it is related. > > I just tested with 1ff1373b3302 (which is the latest commit I checked on > the machine) and the tests do pass... It does show for soft-float. > >> >> FAIL: stdlib/tst-makecontext2 >> FAIL: stdlib/tst-makecontext3 >> FAIL: stdlib/tst-setcontext >> FAIL: stdlib/tst-setcontext2 >> FAIL: stdlib/tst-setcontext4 >> FAIL: stdlib/tst-setcontext7 >> FAIL: stdlib/tst-setcontext9 >> FAIL: stdlib/tst-swapcontext1 >> >> that did not appear in 2.29 testing. This patch restores the removed >> register setting in question, and thus fixes those failures. >> >> Tested for powerpc (soft-float). >> >> 2019-07-30 Joseph Myers <joseph@codesourcery.com> >> >> * sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S >> (__CONTEXT_FUNC_NAME): Restore setting of r31. >> >> diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S b/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S >> index efebb10bba..6fa1ab7d6e 100644 >> --- a/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S >> +++ b/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S >> @@ -276,6 +276,9 @@ ENTRY(__CONTEXT_FUNC_NAME) >> cmpwi r3,0 >> bne 3f /* L(error_exit) */ >> >> + lwz r4,_FRAME_PARM_SAVE2(r1) >> + lwz r31,_UC_REGS_PTR(r4) >> + >> #ifdef __CONTEXT_ENABLE_FPRS >> # ifdef __CONTEXT_ENABLE_VRS >> >>
On 7/30/19 3:17 AM, Florian Weimer wrote: > * Joseph Myers: > >> diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S b/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S >> index efebb10bba..6fa1ab7d6e 100644 >> --- a/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S >> +++ b/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S >> @@ -276,6 +276,9 @@ ENTRY(__CONTEXT_FUNC_NAME) >> cmpwi r3,0 >> bne 3f /* L(error_exit) */ >> >> + lwz r4,_FRAME_PARM_SAVE2(r1) >> + lwz r31,_UC_REGS_PTR(r4) >> + >> #ifdef __CONTEXT_ENABLE_FPRS >> # ifdef __CONTEXT_ENABLE_VRS > > Patch looks good to me. Carlos will have to ack it as the RM, though. Please commit to master. Fixing release regressions like this is the highest priority.
diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S b/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S index efebb10bba..6fa1ab7d6e 100644 --- a/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S +++ b/sysdeps/unix/sysv/linux/powerpc/powerpc32/swapcontext-common.S @@ -276,6 +276,9 @@ ENTRY(__CONTEXT_FUNC_NAME) cmpwi r3,0 bne 3f /* L(error_exit) */ + lwz r4,_FRAME_PARM_SAVE2(r1) + lwz r31,_UC_REGS_PTR(r4) + #ifdef __CONTEXT_ENABLE_FPRS # ifdef __CONTEXT_ENABLE_VRS