This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
PowerPC64 thread-safe stubs not needed for iplt
- From: Alan Modra <amodra at gmail dot com>
- To: binutils at sourceware dot org
- Date: Thu, 26 Feb 2015 21:48:22 +1030
- Subject: PowerPC64 thread-safe stubs not needed for iplt
- Authentication-results: sourceware.org; auth=none
I was looking at a current glibc using objdump today and saw an odd
plt call stub.
0000000000044d80 <00000033.plt_call.__strchrnul>:
44d80: f8 41 00 28 std r2,40(r1)
44d84: e9 82 8c f8 ld r12,-29448(r2)
44d88: 7d 89 03 a6 mtctr r12
44d8c: e8 42 8d 00 ld r2,-29440(r2)
44d90: 28 22 00 00 cmpldi r2,0
44d94: 4c e2 04 20 bnectr+
44d98: 48 13 84 f0 b 17d288 <realloc@plt>
What? It doesn't branch to __strchrnul@plt on finding a zero r2?
Turns out this isn't a real problem since the stub is for loading an
ifunc, so will not be lazily resolved and thus r2 will never be zero.
Of course, that means the thread-safety check is unnecessary.
I also tweak the special __tls_get_addr_opt call stub here, to
restore r2 immediately after the call. Not doing that might affect
eh_frame unwinding.
* elf64-ppc.c (plt_stub_size, build_plt_stub): Don't build
thread-safe stubs for iplt.
(build_tls_get_addr_stub): Restore r2 immediately after call.
diff --git a/bfd/elf64-ppc.c b/bfd/elf64-ppc.c
index 1573e30..90c2cc0 100644
--- a/bfd/elf64-ppc.c
+++ b/bfd/elf64-ppc.c
@@ -10220,7 +10220,10 @@ plt_stub_size (struct ppc_link_hash_table *htab,
size += 4;
if (htab->params->plt_static_chain)
size += 4;
- if (htab->params->plt_thread_safe)
+ if (htab->params->plt_thread_safe
+ && htab->elf.dynamic_sections_created
+ && stub_entry->h != NULL
+ && stub_entry->h->elf.dynindx != -1)
size += 8;
if (PPC_HA (off + 8 + 8 * htab->params->plt_static_chain) != PPC_HA (off))
size += 4;
@@ -10260,16 +10263,18 @@ build_plt_stub (struct ppc_link_hash_table *htab,
bfd *obfd = htab->params->stub_bfd;
bfd_boolean plt_load_toc = htab->opd_abi;
bfd_boolean plt_static_chain = htab->params->plt_static_chain;
- bfd_boolean plt_thread_safe = htab->params->plt_thread_safe;
+ bfd_boolean plt_thread_safe = (htab->params->plt_thread_safe
+ && htab->elf.dynamic_sections_created
+ && stub_entry->h != NULL
+ && stub_entry->h->elf.dynindx != -1);
bfd_boolean use_fake_dep = plt_thread_safe;
bfd_vma cmp_branch_off = 0;
if (!ALWAYS_USE_FAKE_DEP
&& plt_load_toc
&& plt_thread_safe
- && !(stub_entry->h != NULL
- && (stub_entry->h == htab->tls_get_addr_fd
- || stub_entry->h == htab->tls_get_addr)
+ && !((stub_entry->h == htab->tls_get_addr_fd
+ || stub_entry->h == htab->tls_get_addr)
&& !htab->params->no_tls_get_addr_opt))
{
bfd_vma pltoff = stub_entry->plt_ent->plt.offset & ~1;
@@ -10462,8 +10467,8 @@ build_tls_get_addr_stub (struct ppc_link_hash_table *htab,
p = build_plt_stub (htab, stub_entry, p, offset, r);
bfd_put_32 (obfd, BCTRL, p - 4);
- bfd_put_32 (obfd, LD_R11_0R1 + STK_LINKER (htab), p), p += 4;
bfd_put_32 (obfd, LD_R2_0R1 + STK_TOC (htab), p), p += 4;
+ bfd_put_32 (obfd, LD_R11_0R1 + STK_LINKER (htab), p), p += 4;
bfd_put_32 (obfd, MTLR_R11, p), p += 4;
bfd_put_32 (obfd, BLR, p), p += 4;
--
Alan Modra
Australia Development Lab, IBM