[2.43] LoongArch: Fix broken DESC => IE transition for 2.43 branch
Checks
Commit Message
If code compiled with -fPIC -mtls-dialect=desc is linked into a PDE or
PIE, and the code refers to external DSO symbols, we can produce broken
link unit as check_relocs expects DESC => IE transition to happen and
emits a TLS IE entry in the GOT, but a too early "continue" in
relax_section actually jumps over the DESC => IE transition so the code
sequence is unchanged and still expecting a TLS descriptor (instead of
an IE entry) in the GOT.
The bug is already fixed in master branch by commit 5c3d09c1855b
("LoongArch: Optimize the relaxation process") so this fix is only
needed for the 2.43 branch.
Reported-by: Icenowy Zheng <uwu@icenowy.me>
Closes: https://gcc.gnu.org/PR118114
Signed-off-by: Xi Ruoyao <xry111@xry111.site>
---
bfd/elfnn-loongarch.c | 43 ++++++++++++++++++++++---------------------
1 file changed, 22 insertions(+), 21 deletions(-)
base-commit: 834b779ad28c3cecb5ad91a1ab2af3c8453c3b3a
Comments
在 2024-12-19星期四的 13:23 +0800，Xi Ruoyao写道：
> If code compiled with -fPIC -mtls-dialect=desc is linked into a PDE
> or
> PIE, and the code refers to external DSO symbols, we can produce
> broken
> link unit as check_relocs expects DESC => IE transition to happen and
> emits a TLS IE entry in the GOT, but a too early "continue" in
> relax_section actually jumps over the DESC => IE transition so the
> code
> sequence is unchanged and still expecting a TLS descriptor (instead
> of
> an IE entry) in the GOT.
>
> The bug is already fixed in master branch by commit 5c3d09c1855b
> ("LoongArch: Optimize the relaxation process") so this fix is only
> needed for the 2.43 branch.
>
> Reported-by: Icenowy Zheng <uwu@icenowy.me>
My original case (which is compiling a full-featured Mesa) fails in
another way after applying this patch -- originally it fails because
intel_clc gets segfault (which uses std::call_once to initialize LLVM),
now it fails when linking some .so files (including libgallium).
Here's the backtrace:
```
#0 loongarch_can_trans_tls (input_bfd=<optimized out>,
info=0x555558f60b78 <link_info>, h=0x0, r_symndx=<optimized out>,
r_type=<optimized out>)
at /var/cache/acbs/build/acbs._4a9gw_q/binutils-2.43.1/bfd/elfnn-
loongarch.c:787
#1 0x00007ffff01b44fc in loongarch_tls_transition
(input_bfd=0x55558ed785e0, info=0x555558f60b78 <link_info>, h=0x0,
r_symndx=98, r_type=111)
at /var/cache/acbs/build/acbs._4a9gw_q/binutils-2.43.1/bfd/elfnn-
loongarch.c:844
#2 loongarch_elf_check_relocs (abfd=<optimized out>,
info=0x555558f60b78 <link_info>, sec=0x55558e1d71d8, relocs=<optimized
out>)
at /var/cache/acbs/build/acbs._4a9gw_q/binutils-2.43.1/bfd/elfnn-
loongarch.c:985
#3 0x00007ffff01eb828 in _bfd_elf_link_iterate_on_relocs
(abfd=0x55558ed785e0, info=0x555558f60b78 <link_info>,
action=0x7ffff01b4040 <loongarch_elf_check_relocs>)
at /var/cache/acbs/build/acbs._4a9gw_q/binutils-
2.43.1/bfd/elflink.c:4257
#4 0x0000555558ed39c0 in lang_check_relocs () at
/var/cache/acbs/build/acbs._4a9gw_q/binutils-2.43.1/ld/ldlang.c:8071
#5 lang_process () at /var/cache/acbs/build/acbs._4a9gw_q/binutils-
2.43.1/ld/ldlang.c:8383
#6 0x0000555558eb8564 in main (argc=<optimized out>, argv=<optimized
out>) at /var/cache/acbs/build/acbs._4a9gw_q/binutils-
2.43.1/ld/ldmain.c:529
```
> Closes: https://gcc.gnu.org/PR118114
> Signed-off-by: Xi Ruoyao <xry111@xry111.site>
> ---
> bfd/elfnn-loongarch.c | 43 ++++++++++++++++++++++-------------------
> --
> 1 file changed, 22 insertions(+), 21 deletions(-)
>
> diff --git a/bfd/elfnn-loongarch.c b/bfd/elfnn-loongarch.c
> index 312707bb00b..1c34797c2e1 100644
> --- a/bfd/elfnn-loongarch.c
> +++ b/bfd/elfnn-loongarch.c
> @@ -5326,7 +5326,7 @@ loongarch_elf_relax_section (bfd *abfd,
> asection *sec,
> {
> char symtype;
> bfd_vma symval;
> - asection *sym_sec;
> + asection *sym_sec = NULL;
> bool local_got = false;
> Elf_Internal_Rela *rel = relocs + i;
> struct elf_link_hash_entry *h = NULL;
> @@ -5418,14 +5418,33 @@ loongarch_elf_relax_section (bfd *abfd,
> asection *sec,
> symval = h->root.u.def.value;
> sym_sec = h->root.u.def.section;
> }
> - else
> - continue;
>
> if (h && LARCH_REF_LOCAL (info, h))
> local_got = true;
> symtype = h->type;
> }
>
> + /* If the conditions for tls type transition are met, type
> + transition is performed instead of relax.
> + During the transition from DESC->IE/LE, there are 2
> situations
> + depending on the different configurations of the
> relax/norelax
> + option.
> + If the -relax option is used, the extra nops will be
> removed,
> + and this transition is performed in pass 0.
> + If the --no-relax option is used, nop will be retained, and
> + this transition is performed in pass 1. */
> + if (IS_LOONGARCH_TLS_TRANS_RELOC (r_type)
> + && (i + 1 != sec->reloc_count)
> + && ELFNN_R_TYPE (rel[1].r_info) == R_LARCH_RELAX
> + && loongarch_can_trans_tls (abfd, info, h, r_symndx,
> r_type))
> + {
> + loongarch_tls_perform_trans (abfd, sec, rel, h, info);
> + r_type = ELFNN_R_TYPE (rel->r_info);
> + }
> +
> + if (!sym_sec)
> + continue;
> +
> if (sym_sec->sec_info_type == SEC_INFO_TYPE_MERGE
> && (sym_sec->flags & SEC_MERGE))
> {
> @@ -5453,24 +5472,6 @@ loongarch_elf_relax_section (bfd *abfd,
> asection *sec,
>
> symval += sec_addr (sym_sec);
>
> - /* If the conditions for tls type transition are met, type
> - transition is performed instead of relax.
> - During the transition from DESC->IE/LE, there are 2
> situations
> - depending on the different configurations of the
> relax/norelax
> - option.
> - If the -relax option is used, the extra nops will be
> removed,
> - and this transition is performed in pass 0.
> - If the --no-relax option is used, nop will be retained, and
> - this transition is performed in pass 1. */
> - if (IS_LOONGARCH_TLS_TRANS_RELOC (r_type)
> - && (i + 1 != sec->reloc_count)
> - && ELFNN_R_TYPE (rel[1].r_info) == R_LARCH_RELAX
> - && loongarch_can_trans_tls (abfd, info, h, r_symndx,
> r_type))
> - {
> - loongarch_tls_perform_trans (abfd, sec, rel, h, info);
> - r_type = ELFNN_R_TYPE (rel->r_info);
> - }
> -
> switch (r_type)
> {
> case R_LARCH_ALIGN:
>
> base-commit: 834b779ad28c3cecb5ad91a1ab2af3c8453c3b3a
On Thu, 2024-12-19 at 17:04 +0800, Icenowy Zheng wrote:
> 在 2024-12-19星期四的 13:23 +0800，Xi Ruoyao写道：
> > If code compiled with -fPIC -mtls-dialect=desc is linked into a PDE
> > or
> > PIE, and the code refers to external DSO symbols, we can produce
> > broken
> > link unit as check_relocs expects DESC => IE transition to happen and
> > emits a TLS IE entry in the GOT, but a too early "continue" in
> > relax_section actually jumps over the DESC => IE transition so the
> > code
> > sequence is unchanged and still expecting a TLS descriptor (instead
> > of
> > an IE entry) in the GOT.
> >
> > The bug is already fixed in master branch by commit 5c3d09c1855b
> > ("LoongArch: Optimize the relaxation process") so this fix is only
> > needed for the 2.43 branch.
> >
> > Reported-by: Icenowy Zheng <uwu@icenowy.me>
>
> My original case (which is compiling a full-featured Mesa) fails in
> another way after applying this patch -- originally it fails because
> intel_clc gets segfault (which uses std::call_once to initialize LLVM),
> now it fails when linking some .so files (including libgallium).
>
> Here's the backtrace:
>
> ```
> #0 loongarch_can_trans_tls (input_bfd=<optimized out>,
> info=0x555558f60b78 <link_info>, h=0x0, r_symndx=<optimized out>,
> r_type=<optimized out>)
> at /var/cache/acbs/build/acbs._4a9gw_q/binutils-2.43.1/bfd/elfnn-
> loongarch.c:787
I guess (not sure) it's fixed by
commit 28489a70d4660d67e71d75e82286a6e1a7003b93
Author: Xin Wang <yw987194828@gmail.com>
Date: Fri Sep 6 09:00:12 2024 +0800
LoongArch: Add elfNN_loongarch_mkobject to initialize LoongArch tdata
This is already backported to 2.43 branch as commit
c40337b1784d81c615a5dd9fd391f6677e52c6d8.
在 2024-12-19星期四的 17:18 +0800，Xi Ruoyao写道：
> On Thu, 2024-12-19 at 17:04 +0800, Icenowy Zheng wrote:
> > 在 2024-12-19星期四的 13:23 +0800，Xi Ruoyao写道：
> > > If code compiled with -fPIC -mtls-dialect=desc is linked into a
> > > PDE
> > > or
> > > PIE, and the code refers to external DSO symbols, we can produce
> > > broken
> > > link unit as check_relocs expects DESC => IE transition to happen
> > > and
> > > emits a TLS IE entry in the GOT, but a too early "continue" in
> > > relax_section actually jumps over the DESC => IE transition so
> > > the
> > > code
> > > sequence is unchanged and still expecting a TLS descriptor
> > > (instead
> > > of
> > > an IE entry) in the GOT.
> > >
> > > The bug is already fixed in master branch by commit 5c3d09c1855b
> > > ("LoongArch: Optimize the relaxation process") so this fix is
> > > only
> > > needed for the 2.43 branch.
> > >
> > > Reported-by: Icenowy Zheng <uwu@icenowy.me>
> >
> > My original case (which is compiling a full-featured Mesa) fails in
> > another way after applying this patch -- originally it fails
> > because
> > intel_clc gets segfault (which uses std::call_once to initialize
> > LLVM),
> > now it fails when linking some .so files (including libgallium).
> >
> > Here's the backtrace:
> >
> > ```
> > #0 loongarch_can_trans_tls (input_bfd=<optimized out>,
> > info=0x555558f60b78 <link_info>, h=0x0, r_symndx=<optimized out>,
> > r_type=<optimized out>)
> > at /var/cache/acbs/build/acbs._4a9gw_q/binutils-
> > 2.43.1/bfd/elfnn-
> > loongarch.c:787
>
> I guess (not sure) it's fixed by
>
> commit 28489a70d4660d67e71d75e82286a6e1a7003b93
> Author: Xin Wang <yw987194828@gmail.com>
> Date: Fri Sep 6 09:00:12 2024 +0800
>
> LoongArch: Add elfNN_loongarch_mkobject to initialize LoongArch
> tdata
>
Yes, picking this patch makes everything normal now.
So, for the original patch in the mail's subject,
Tested-By: Icenowy Zheng <uwu@icenowy.me>
Thanks!
Icenowy
> This is already backported to 2.43 branch as commit
> c40337b1784d81c615a5dd9fd391f6677e52c6d8.
>
When I apply this patch on the 2.43 branch and compile with gcc 15, the
following warning is triggered.
binutils-gdb/bfd/elfnn-loongarch.c: In function
‘loongarch_elf_relax_section’:
binutils-gdb/bfd/elfnn-loongarch.c:5471:16: error: ‘symval’ may be used
uninitialized [-Werror=maybe-uninitialized]
5471 | symval += rel->r_addend;
| ~~~~~~~^~~~~~~~~~~~~~~~~
binutils-gdb/bfd/elfnn-loongarch.c:5328:15: note: ‘symval’ was declared here
5328 | bfd_vma symval;
| ^~~~~~
cc1: all warnings being treated as errors
On 12/19/24 1:23 PM, Xi Ruoyao wrote:
> If code compiled with -fPIC -mtls-dialect=desc is linked into a PDE or
> PIE, and the code refers to external DSO symbols, we can produce broken
> link unit as check_relocs expects DESC => IE transition to happen and
> emits a TLS IE entry in the GOT, but a too early "continue" in
> relax_section actually jumps over the DESC => IE transition so the code
> sequence is unchanged and still expecting a TLS descriptor (instead of
> an IE entry) in the GOT.
>
> The bug is already fixed in master branch by commit 5c3d09c1855b
> ("LoongArch: Optimize the relaxation process") so this fix is only
> needed for the 2.43 branch.
>
> Reported-by: Icenowy Zheng <uwu@icenowy.me>
> Closes: https://gcc.gnu.org/PR118114
> Signed-off-by: Xi Ruoyao <xry111@xry111.site>
> ---
> bfd/elfnn-loongarch.c | 43 ++++++++++++++++++++++---------------------
> 1 file changed, 22 insertions(+), 21 deletions(-)
>
> diff --git a/bfd/elfnn-loongarch.c b/bfd/elfnn-loongarch.c
> index 312707bb00b..1c34797c2e1 100644
> --- a/bfd/elfnn-loongarch.c
> +++ b/bfd/elfnn-loongarch.c
> @@ -5326,7 +5326,7 @@ loongarch_elf_relax_section (bfd *abfd, asection *sec,
> {
> char symtype;
> bfd_vma symval;
> - asection *sym_sec;
> + asection *sym_sec = NULL;
> bool local_got = false;
> Elf_Internal_Rela *rel = relocs + i;
> struct elf_link_hash_entry *h = NULL;
> @@ -5418,14 +5418,33 @@ loongarch_elf_relax_section (bfd *abfd, asection *sec,
> symval = h->root.u.def.value;
> sym_sec = h->root.u.def.section;
> }
> - else
> - continue;
>
> if (h && LARCH_REF_LOCAL (info, h))
> local_got = true;
> symtype = h->type;
> }
>
> + /* If the conditions for tls type transition are met, type
> + transition is performed instead of relax.
> + During the transition from DESC->IE/LE, there are 2 situations
> + depending on the different configurations of the relax/norelax
> + option.
> + If the -relax option is used, the extra nops will be removed,
> + and this transition is performed in pass 0.
> + If the --no-relax option is used, nop will be retained, and
> + this transition is performed in pass 1. */
> + if (IS_LOONGARCH_TLS_TRANS_RELOC (r_type)
> + && (i + 1 != sec->reloc_count)
> + && ELFNN_R_TYPE (rel[1].r_info) == R_LARCH_RELAX
> + && loongarch_can_trans_tls (abfd, info, h, r_symndx, r_type))
> + {
> + loongarch_tls_perform_trans (abfd, sec, rel, h, info);
> + r_type = ELFNN_R_TYPE (rel->r_info);
> + }
> +
> + if (!sym_sec)
> + continue;
> +
> if (sym_sec->sec_info_type == SEC_INFO_TYPE_MERGE
> && (sym_sec->flags & SEC_MERGE))
> {
> @@ -5453,24 +5472,6 @@ loongarch_elf_relax_section (bfd *abfd, asection *sec,
>
> symval += sec_addr (sym_sec);
>
> - /* If the conditions for tls type transition are met, type
> - transition is performed instead of relax.
> - During the transition from DESC->IE/LE, there are 2 situations
> - depending on the different configurations of the relax/norelax
> - option.
> - If the -relax option is used, the extra nops will be removed,
> - and this transition is performed in pass 0.
> - If the --no-relax option is used, nop will be retained, and
> - this transition is performed in pass 1. */
> - if (IS_LOONGARCH_TLS_TRANS_RELOC (r_type)
> - && (i + 1 != sec->reloc_count)
> - && ELFNN_R_TYPE (rel[1].r_info) == R_LARCH_RELAX
> - && loongarch_can_trans_tls (abfd, info, h, r_symndx, r_type))
> - {
> - loongarch_tls_perform_trans (abfd, sec, rel, h, info);
> - r_type = ELFNN_R_TYPE (rel->r_info);
> - }
> -
> switch (r_type)
> {
> case R_LARCH_ALIGN:
>
> base-commit: 834b779ad28c3cecb5ad91a1ab2af3c8453c3b3a
On Fri, 2024-12-27 at 14:12 +0800, Lulu Cai wrote:
> When I apply this patch on the 2.43 branch and compile with gcc 15, the
> following warning is triggered.
>
> binutils-gdb/bfd/elfnn-loongarch.c: In function
> ‘loongarch_elf_relax_section’:
> binutils-gdb/bfd/elfnn-loongarch.c:5471:16: error: ‘symval’ may be used
> uninitialized [-Werror=maybe-uninitialized]
> 5471 | symval += rel->r_addend;
> > ~~~~~~~^~~~~~~~~~~~~~~~~
> binutils-gdb/bfd/elfnn-loongarch.c:5328:15: note: ‘symval’ was declared here
> 5328 | bfd_vma symval;
> > ^~~~~~
> cc1: all warnings being treated as errors
Hmm it seems a compiler bug... Whenever we set sym_sec to non-NULL,
symval is also set. And line 5471 should be skipped with the "continue"
at line 5446 then if sym_sec is NULL.
I'll initialize the variable in v2 but also send a GCC bug report.
> On 12/19/24 1:23 PM, Xi Ruoyao wrote:
> > If code compiled with -fPIC -mtls-dialect=desc is linked into a PDE or
> > PIE, and the code refers to external DSO symbols, we can produce broken
> > link unit as check_relocs expects DESC => IE transition to happen and
> > emits a TLS IE entry in the GOT, but a too early "continue" in
> > relax_section actually jumps over the DESC => IE transition so the code
> > sequence is unchanged and still expecting a TLS descriptor (instead of
> > an IE entry) in the GOT.
> >
> > The bug is already fixed in master branch by commit 5c3d09c1855b
> > ("LoongArch: Optimize the relaxation process") so this fix is only
> > needed for the 2.43 branch.
> >
> > Reported-by: Icenowy Zheng <uwu@icenowy.me>
> > Closes: https://gcc.gnu.org/PR118114
> > Signed-off-by: Xi Ruoyao <xry111@xry111.site>
> > ---
> > bfd/elfnn-loongarch.c | 43 ++++++++++++++++++++++---------------------
> > 1 file changed, 22 insertions(+), 21 deletions(-)
> >
> > diff --git a/bfd/elfnn-loongarch.c b/bfd/elfnn-loongarch.c
> > index 312707bb00b..1c34797c2e1 100644
> > --- a/bfd/elfnn-loongarch.c
> > +++ b/bfd/elfnn-loongarch.c
> > @@ -5326,7 +5326,7 @@ loongarch_elf_relax_section (bfd *abfd, asection *sec,
> > {
> > char symtype;
> > bfd_vma symval;
> > - asection *sym_sec;
> > + asection *sym_sec = NULL;
> > bool local_got = false;
> > Elf_Internal_Rela *rel = relocs + i;
> > struct elf_link_hash_entry *h = NULL;
> > @@ -5418,14 +5418,33 @@ loongarch_elf_relax_section (bfd *abfd, asection *sec,
> > symval = h->root.u.def.value;
> > sym_sec = h->root.u.def.section;
> > }
> > - else
> > - continue;
> >
> > if (h && LARCH_REF_LOCAL (info, h))
> > local_got = true;
> > symtype = h->type;
> > }
> >
> > + /* If the conditions for tls type transition are met, type
> > + transition is performed instead of relax.
> > + During the transition from DESC->IE/LE, there are 2 situations
> > + depending on the different configurations of the relax/norelax
> > + option.
> > + If the -relax option is used, the extra nops will be removed,
> > + and this transition is performed in pass 0.
> > + If the --no-relax option is used, nop will be retained, and
> > + this transition is performed in pass 1. */
> > + if (IS_LOONGARCH_TLS_TRANS_RELOC (r_type)
> > + && (i + 1 != sec->reloc_count)
> > + && ELFNN_R_TYPE (rel[1].r_info) == R_LARCH_RELAX
> > + && loongarch_can_trans_tls (abfd, info, h, r_symndx, r_type))
> > + {
> > + loongarch_tls_perform_trans (abfd, sec, rel, h, info);
> > + r_type = ELFNN_R_TYPE (rel->r_info);
> > + }
> > +
> > + if (!sym_sec)
> > + continue;
> > +
> > if (sym_sec->sec_info_type == SEC_INFO_TYPE_MERGE
> > && (sym_sec->flags & SEC_MERGE))
> > {
> > @@ -5453,24 +5472,6 @@ loongarch_elf_relax_section (bfd *abfd, asection *sec,
> >
> > symval += sec_addr (sym_sec);
> >
> > - /* If the conditions for tls type transition are met, type
> > - transition is performed instead of relax.
> > - During the transition from DESC->IE/LE, there are 2 situations
> > - depending on the different configurations of the relax/norelax
> > - option.
> > - If the -relax option is used, the extra nops will be removed,
> > - and this transition is performed in pass 0.
> > - If the --no-relax option is used, nop will be retained, and
> > - this transition is performed in pass 1. */
> > - if (IS_LOONGARCH_TLS_TRANS_RELOC (r_type)
> > - && (i + 1 != sec->reloc_count)
> > - && ELFNN_R_TYPE (rel[1].r_info) == R_LARCH_RELAX
> > - && loongarch_can_trans_tls (abfd, info, h, r_symndx, r_type))
> > - {
> > - loongarch_tls_perform_trans (abfd, sec, rel, h, info);
> > - r_type = ELFNN_R_TYPE (rel->r_info);
> > - }
> > -
> > switch (r_type)
> > {
> > case R_LARCH_ALIGN:
> >
> > base-commit: 834b779ad28c3cecb5ad91a1ab2af3c8453c3b3a
>
>
@@ -5326,7 +5326,7 @@ loongarch_elf_relax_section (bfd *abfd, asection *sec,
{
char symtype;
bfd_vma symval;
- asection *sym_sec;
+ asection *sym_sec = NULL;
bool local_got = false;
Elf_Internal_Rela *rel = relocs + i;
struct elf_link_hash_entry *h = NULL;
@@ -5418,14 +5418,33 @@ loongarch_elf_relax_section (bfd *abfd, asection *sec,
symval = h->root.u.def.value;
sym_sec = h->root.u.def.section;
}
- else
- continue;
if (h && LARCH_REF_LOCAL (info, h))
local_got = true;
symtype = h->type;
}
+ /* If the conditions for tls type transition are met, type
+ transition is performed instead of relax.
+ During the transition from DESC->IE/LE, there are 2 situations
+ depending on the different configurations of the relax/norelax
+ option.
+ If the -relax option is used, the extra nops will be removed,
+ and this transition is performed in pass 0.
+ If the --no-relax option is used, nop will be retained, and
+ this transition is performed in pass 1. */
+ if (IS_LOONGARCH_TLS_TRANS_RELOC (r_type)
+ && (i + 1 != sec->reloc_count)
+ && ELFNN_R_TYPE (rel[1].r_info) == R_LARCH_RELAX
+ && loongarch_can_trans_tls (abfd, info, h, r_symndx, r_type))
+ {
+ loongarch_tls_perform_trans (abfd, sec, rel, h, info);
+ r_type = ELFNN_R_TYPE (rel->r_info);
+ }
+
+ if (!sym_sec)
+ continue;
+
if (sym_sec->sec_info_type == SEC_INFO_TYPE_MERGE
&& (sym_sec->flags & SEC_MERGE))
{
@@ -5453,24 +5472,6 @@ loongarch_elf_relax_section (bfd *abfd, asection *sec,
symval += sec_addr (sym_sec);
- /* If the conditions for tls type transition are met, type
- transition is performed instead of relax.
- During the transition from DESC->IE/LE, there are 2 situations
- depending on the different configurations of the relax/norelax
- option.
- If the -relax option is used, the extra nops will be removed,
- and this transition is performed in pass 0.
- If the --no-relax option is used, nop will be retained, and
- this transition is performed in pass 1. */
- if (IS_LOONGARCH_TLS_TRANS_RELOC (r_type)
- && (i + 1 != sec->reloc_count)
- && ELFNN_R_TYPE (rel[1].r_info) == R_LARCH_RELAX
- && loongarch_can_trans_tls (abfd, info, h, r_symndx, r_type))
- {
- loongarch_tls_perform_trans (abfd, sec, rel, h, info);
- r_type = ELFNN_R_TYPE (rel->r_info);
- }
-
switch (r_type)
{
case R_LARCH_ALIGN: