* The ltoff22x relaxation bug on ia64.
@ 2003-03-17 18:47 H. J. Lu
2003-03-18 21:02 ` [PATCH] Fix " Jakub Jelinek
0 siblings, 1 reply; 3+ messages in thread
From: H. J. Lu @ 2003-03-17 18:47 UTC (permalink / raw)
To: Jakub Jelinek; +Cc: rth, binutils
On Mon, Mar 17, 2003 at 09:10:57AM -0800, H. J. Lu wrote:
>
> There is another ia64 linker bug. I can't build the current
> gcc-3_2-rhl8-branch in CVS using the current binutils on Linux/ia64. I
> got
>
> ../boehm-gc/.libs/libgcjgc_convenience.a(reclaim.o)(.text+0xd1): In function
> `report_leak':
> /net/gnu/export/gnu/src/gcc-3.2-redhat-8/gcc/boehm-gc/reclaim.c:40: relocation
> truncated to fit: GPREL22 .text
> collect2: ld returned 1 exit status
> make[5]: *** [libgcj.la] Error 1
> make[5]: Leaving directory
> `/export/build/gnu/gcc-3.2-redhat-8/build-ia64-linux/ia64-unknown-linux-gnu/libjava'
> m
>
> It is a new bug.
>
The problem is
http://sources.redhat.com/ml/binutils/2003-02/msg00478.html
doesn't take SHF_MERGE into account. The symaddr computed in
elfNN_ia64_relax_section may be different from the one used.
H.J.
^ permalink raw reply [flat|nested] 3+ messages in thread
* [PATCH] Fix ltoff22x relaxation bug on ia64.
2003-03-17 18:47 The ltoff22x relaxation bug on ia64 H. J. Lu
@ 2003-03-18 21:02 ` Jakub Jelinek
2003-03-18 22:40 ` Richard Henderson
0 siblings, 1 reply; 3+ messages in thread
From: Jakub Jelinek @ 2003-03-18 21:02 UTC (permalink / raw)
To: Richard Henderson, H. J. Lu; +Cc: binutils
On Mon, Mar 17, 2003 at 10:47:49AM -0800, H. J. Lu wrote:
> On Mon, Mar 17, 2003 at 09:10:57AM -0800, H. J. Lu wrote:
> >
> > There is another ia64 linker bug. I can't build the current
> > gcc-3_2-rhl8-branch in CVS using the current binutils on Linux/ia64. I
> > got
> >
> > ../boehm-gc/.libs/libgcjgc_convenience.a(reclaim.o)(.text+0xd1): In function
> > `report_leak':
> > /net/gnu/export/gnu/src/gcc-3.2-redhat-8/gcc/boehm-gc/reclaim.c:40: relocation
> > truncated to fit: GPREL22 .text
> > collect2: ld returned 1 exit status
> > make[5]: *** [libgcj.la] Error 1
> > make[5]: Leaving directory
> > `/export/build/gnu/gcc-3.2-redhat-8/build-ia64-linux/ia64-unknown-linux-gnu/libjava'
> > m
> >
> > It is a new bug.
> >
>
> The problem is
>
> http://sources.redhat.com/ml/binutils/2003-02/msg00478.html
>
> doesn't take SHF_MERGE into account. The symaddr computed in
> elfNN_ia64_relax_section may be different from the one used.
Simplified testcase below, fixed by following patch.
This patch includes separately posted "[PATCH] Fix bug in ia64 trampoline"
patch, although done differently.
Ok to commit?
2003-03-18 Jakub Jelinek <jakub@redhat.com>
* elfxx-ia64.c (elfNN_ia64_relax_section): Handle relaxation
againt mergeable sections. Take r_addend into account when caching
trampolines.
--- bfd/elfxx-ia64.c.jj 2003-03-18 15:18:23.000000000 -0500
+++ bfd/elfxx-ia64.c 2003-03-18 15:47:21.000000000 -0500
@@ -819,6 +819,7 @@ elfNN_ia64_relax_section (abfd, sec, lin
tsec = ia64_info->plt_sec;
toff = dyn_i->plt2_offset;
+ BFD_ASSERT (irel->r_addend == 0);
}
/* Can't do anything else with dynamic symbols. */
@@ -837,10 +838,15 @@ elfNN_ia64_relax_section (abfd, sec, lin
}
}
- symaddr = (tsec->output_section->vma
- + tsec->output_offset
- + toff
- + irel->r_addend);
+ if (tsec->sec_info_type == ELF_INFO_TYPE_MERGE)
+ toff = _bfd_merged_section_offset (abfd, &tsec,
+ elf_section_data (tsec)->sec_info,
+ toff + irel->r_addend,
+ (bfd_vma) 0);
+ else
+ toff += irel->r_addend;
+
+ symaddr = tsec->output_section->vma + tsec->output_offset + toff;
roff = irel->r_offset;
cat > a.s <<EOF
.text
.globl fn1
fn1:
addl r15 = @ltoff(foostr), gp;; ld8 r16 = [r15];;
addl r15 = @ltoff(barstr), gp;; ld8 r16 = [r15];;
addl r15 = @ltoff(bazstr), gp;; ld8 r16 = [r15];;
.section .rodata.str1.8, "aMS", @progbits, 1
.align 8
foostr: .string "foo"
.align 8
.globl barstr
barstr: .string "bar"
.align 8
bazstr: .fill 0x200000-1, 1, 32
.byte 0
EOF
cat > b.s <<EOF
.text
.globl fn2
fn2:
addl r15 = @ltoffx(foost2), gp;; ld8.mov r16 = [r15], foost2;;
addl r15 = @ltoffx(barst2), gp;; ld8.mov r16 = [r15], barst2;;
addl r15 = @ltoffx(bazst2), gp;; ld8.mov r16 = [r15], bazst2;;
addl r15 = @ltoffx(.Lstr2), gp;; ld8.mov r16 = [r15], .Lstr2;;
.section .rodata.str1.8, "aMS", @progbits, 1
.align 8
foost2: .string "foo"
.align 8
.globl barst2
barst2: .string "bar"
.align 8
.globl bazst2
bazst2: .string "baz"
.align 8
.Lstr2: .string "str"
.data
.globl __gp
__gp:
EOF
as -o a.o a.s
as -o b.o b.s
ld -shared -o a.so a.o b.o
Jakub
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] Fix ltoff22x relaxation bug on ia64.
2003-03-18 21:02 ` [PATCH] Fix " Jakub Jelinek
@ 2003-03-18 22:40 ` Richard Henderson
0 siblings, 0 replies; 3+ messages in thread
From: Richard Henderson @ 2003-03-18 22:40 UTC (permalink / raw)
To: Jakub Jelinek; +Cc: H. J. Lu, binutils
On Tue, Mar 18, 2003 at 10:02:53PM +0100, Jakub Jelinek wrote:
> * elfxx-ia64.c (elfNN_ia64_relax_section): Handle relaxation
> againt mergeable sections. Take r_addend into account when caching
> trampolines.
Ok.
Thanks for fixing this.
r~
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2003-03-18 22:40 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2003-03-17 18:47 The ltoff22x relaxation bug on ia64 H. J. Lu
2003-03-18 21:02 ` [PATCH] Fix " Jakub Jelinek
2003-03-18 22:40 ` Richard Henderson
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).