| Message ID | 52D700F6.2060609@redhat.com |
|---|---|
| State | New |
| Headers | show |
On Wed, Jan 15, 2014 at 01:43:18PM -0800, Richard Henderson wrote: > --- /dev/null > +++ b/gcc/testsuite/gcc.target/i386/pr54694.c > @@ -0,0 +1,11 @@ > +/* { dg-do compile } */ > +/* { dg-options "-O" } */ > + > +register void *hfp __asm__("%ebp"); /* { dg-message "note: for" } */ Shouldn't that be %rbp for x86_64? Or do we treat it the same? > +extern void g(void *); > + > +void f(int x) /* { dg-error "frame pointer required" } */ > +{ > + g(__builtin_alloca(x)); > +} Jakub
On Wed, Jan 15, 2014 at 01:43:18PM -0800, Richard Henderson wrote: > On 01/15/2014 08:37 AM, H.J. Lu wrote: > > We should add a testcase to verify this. > > > > I included the following testcase with the commit. I couldn't find a way > to test this properly generically, so I just went with the obvious i386 test. Seems like the test was missing a ChangeLog entry. I've fixed it up. Marek
On 01/15/2014 01:58 PM, Jakub Jelinek wrote: > On Wed, Jan 15, 2014 at 01:43:18PM -0800, Richard Henderson wrote: >> --- /dev/null >> +++ b/gcc/testsuite/gcc.target/i386/pr54694.c >> @@ -0,0 +1,11 @@ >> +/* { dg-do compile } */ >> +/* { dg-options "-O" } */ >> + >> +register void *hfp __asm__("%ebp"); /* { dg-message "note: for" } */ > > Shouldn't that be %rbp for x86_64? Or do we treat it the same? We treat it the same. r~
On Thu, Jan 16, 2014 at 08:06:07AM -0800, Richard Henderson wrote: > On 01/15/2014 01:58 PM, Jakub Jelinek wrote: > > On Wed, Jan 15, 2014 at 01:43:18PM -0800, Richard Henderson wrote: > >> --- /dev/null > >> +++ b/gcc/testsuite/gcc.target/i386/pr54694.c > >> @@ -0,0 +1,11 @@ > >> +/* { dg-do compile } */ > >> +/* { dg-options "-O" } */ > >> + > >> +register void *hfp __asm__("%ebp"); /* { dg-message "note: for" } */ > > > > Shouldn't that be %rbp for x86_64? Or do we treat it the same? > > We treat it the same. BTW, your fix broke the gcc.target/i386/pr9771-1.c test on i686-linux, the problem is that main normally dynamically realigns the stack. Wonder if the test should be turned into dg-do compile, or perhaps a hack like: int xmain() __asm__ ("main"); int xmain() instead of int main() to avoid the dynamic stack realigning in main (limit the test to *linux* then?), supply main written in assembly, something else? Jakub
On Thu, Jan 16, 2014 at 9:35 AM, Jakub Jelinek <jakub@redhat.com> wrote: > On Thu, Jan 16, 2014 at 08:06:07AM -0800, Richard Henderson wrote: >> On 01/15/2014 01:58 PM, Jakub Jelinek wrote: >> > On Wed, Jan 15, 2014 at 01:43:18PM -0800, Richard Henderson wrote: >> >> --- /dev/null >> >> +++ b/gcc/testsuite/gcc.target/i386/pr54694.c >> >> @@ -0,0 +1,11 @@ >> >> +/* { dg-do compile } */ >> >> +/* { dg-options "-O" } */ >> >> + >> >> +register void *hfp __asm__("%ebp"); /* { dg-message "note: for" } */ >> > >> > Shouldn't that be %rbp for x86_64? Or do we treat it the same? >> >> We treat it the same. > > BTW, your fix broke the gcc.target/i386/pr9771-1.c > test on i686-linux, the problem is that main normally dynamically realigns > the stack. Wonder if the test should be turned into dg-do compile, > or perhaps a hack like: > int xmain() __asm__ ("main"); > int xmain() > instead of > int main() > to avoid the dynamic stack realigning in main (limit the test to *linux* > then?), supply main written in assembly, something else? > > Jakub gcc.target/i386/pr9771-1.c has register long *B asm ("ebp"); It won't work without -maccumulate-outgoing-args.
On Thu, Jan 16, 2014 at 09:40:33AM -0800, H.J. Lu wrote: > On Thu, Jan 16, 2014 at 9:35 AM, Jakub Jelinek <jakub@redhat.com> wrote: > > On Thu, Jan 16, 2014 at 08:06:07AM -0800, Richard Henderson wrote: > >> On 01/15/2014 01:58 PM, Jakub Jelinek wrote: > >> > On Wed, Jan 15, 2014 at 01:43:18PM -0800, Richard Henderson wrote: > >> >> --- /dev/null > >> >> +++ b/gcc/testsuite/gcc.target/i386/pr54694.c > >> >> @@ -0,0 +1,11 @@ > >> >> +/* { dg-do compile } */ > >> >> +/* { dg-options "-O" } */ > >> >> + > >> >> +register void *hfp __asm__("%ebp"); /* { dg-message "note: for" } */ > >> > > >> > Shouldn't that be %rbp for x86_64? Or do we treat it the same? > >> > >> We treat it the same. > > > > BTW, your fix broke the gcc.target/i386/pr9771-1.c > > test on i686-linux, the problem is that main normally dynamically realigns > > the stack. Wonder if the test should be turned into dg-do compile, > > or perhaps a hack like: > > int xmain() __asm__ ("main"); > > int xmain() > > instead of > > int main() > > to avoid the dynamic stack realigning in main (limit the test to *linux* > > then?), supply main written in assembly, something else? > > > > Jakub > > gcc.target/i386/pr9771-1.c has > > register long *B asm ("ebp"); > > It won't work without -maccumulate-outgoing-args. Actually it does, with the above written hack for main it works all of: -m32 -O2 -fomit-frame-pointer -ffixed-ebp -m32 -O2 -fomit-frame-pointer -ffixed-ebp -maccumulate-outgoing-args -m32 -O2 -fomit-frame-pointer -ffixed-ebp -mno-accumulate-outgoing-args Jakub
On 01/16/2014 09:35 AM, Jakub Jelinek wrote: > Wonder if the test should be turned into dg-do compile, > or perhaps a hack like: > int xmain() __asm__ ("main"); > int xmain() > instead of > int main() > to avoid the dynamic stack realigning in main (limit the test to *linux* > then?), supply main written in assembly, something else? The __asm__ hack seems reasonable, although you'll also have to deal with __USER_LABEL_PREFIX__. r~
diff --git a/gcc/ChangeLog b/gcc/ChangeLog index c93bf23..075582a 100644 --- a/gcc/ChangeLog +++ b/gcc/ChangeLog @@ -1,3 +1,11 @@ +2014-01-15 Richard Henderson <rth@redhat.com> + + PR debug/54694 + * reginfo.c (global_regs_decl): Globalize. + * rtl.h (global_regs_decl): Declare. + * ira.c (do_reload): Diagnose frame_pointer_needed and it + reserved via global_regs. + 2014-01-15 Teresa Johnson <tejohnson@google.com> * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef. diff --git a/gcc/ira.c b/gcc/ira.c index 41e05f4..ee6010a 100644 --- a/gcc/ira.c +++ b/gcc/ira.c @@ -5532,6 +5532,18 @@ do_reload (void) if (need_dce && optimize) run_fast_dce (); + /* Diagnose uses of the hard frame pointer when it is used as a global + register. Often we can get away with letting the user appropriate + the frame pointer, but we should let them know when code generation + makes that impossible. */ + if (global_regs[HARD_FRAME_POINTER_REGNUM] && frame_pointer_needed) + { + tree decl = global_regs_decl[HARD_FRAME_POINTER_REGNUM]; + error_at (DECL_SOURCE_LOCATION (current_function_decl), + "frame pointer required, but reserved"); + inform (DECL_SOURCE_LOCATION (decl), "for %qD", decl); + } + timevar_pop (TV_IRA); } diff --git a/gcc/reginfo.c b/gcc/reginfo.c index efaa0cb..bdb980d 100644 --- a/gcc/reginfo.c +++ b/gcc/reginfo.c @@ -86,7 +86,7 @@ static const char initial_call_really_used_regs[] = CALL_REALLY_USED_REGISTERS; char global_regs[FIRST_PSEUDO_REGISTER]; /* Declaration for the global register. */ -static tree GTY(()) global_regs_decl[FIRST_PSEUDO_REGISTER]; +tree global_regs_decl[FIRST_PSEUDO_REGISTER]; /* Same information as REGS_INVALIDATED_BY_CALL but in regset form to be used in dataflow more conveniently. */ diff --git a/gcc/rtl.h b/gcc/rtl.h index e7d60ee..10ee818 100644 --- a/gcc/rtl.h +++ b/gcc/rtl.h @@ -2795,6 +2795,8 @@ extern void _fatal_insn (const char *, const_rtx, const char *, int, const char #define fatal_insn_not_found(insn) \ _fatal_insn_not_found (insn, __FILE__, __LINE__, __FUNCTION__) +/* reginfo.c */ +extern tree GTY(()) global_regs_decl[FIRST_PSEUDO_REGISTER]; #endif /* ! GCC_RTL_H */ diff --git a/gcc/testsuite/gcc.target/i386/pr54694.c b/gcc/testsuite/gcc.target/i386/pr54694.c new file mode 100644 index 0000000..bcf82c2 --- /dev/null +++ b/gcc/testsuite/gcc.target/i386/pr54694.c @@ -0,0 +1,11 @@ +/* { dg-do compile } */ +/* { dg-options "-O" } */ + +register void *hfp __asm__("%ebp"); /* { dg-message "note: for" } */ + +extern void g(void *); + +void f(int x) /* { dg-error "frame pointer required" } */ +{ + g(__builtin_alloca(x)); +}