From: Alan Modra <amodra@sourceware.org>
To: bfd-cvs@sourceware.org
Subject: [binutils-gdb] PR29189, dlltool delaylibs corrupt float/double arguments
Date: Thu, 25 May 2023 07:44:45 +0000 (GMT) [thread overview]
Message-ID: <20230525074445.CA4923858C1F@sourceware.org> (raw)
https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=7529ff1fcdbe260a0ac84ee8f33f4fa4ee1ac455
commit 7529ff1fcdbe260a0ac84ee8f33f4fa4ee1ac455
Author: Alan Modra <amodra@gmail.com>
Date: Mon May 15 10:44:29 2023 +0930
PR29189, dlltool delaylibs corrupt float/double arguments
PR 29189
* dlltool.c (i386_x64_trampoline): Save and restore xmm0-5. Make
use of parameter save area for integer arg regs. Comment.
Diff:
---
binutils/dlltool.c | 52 +++++++++++++++++++++++++++++++++++++++-------------
1 file changed, 39 insertions(+), 13 deletions(-)
diff --git a/binutils/dlltool.c b/binutils/dlltool.c
index 31c864d7d5c..142d43aff81 100644
--- a/binutils/dlltool.c
+++ b/binutils/dlltool.c
@@ -583,22 +583,48 @@ static const char i386_trampoline[] =
"\tpopl %%ecx\n"
"\tjmp *%%eax\n";
+/* Save integer arg regs in parameter space reserved by our caller
+ above the return address. Allocate space for six fp arg regs plus
+ parameter space possibly used by __delayLoadHelper2 plus alignment.
+ We enter with the stack offset from 16-byte alignment by the return
+ address, so allocate 96 + 32 + 8 = 136 bytes. Note that only the
+ first four xmm regs are used to pass fp args, but the first six
+ vector ymm (zmm too?) are used to pass vector args. We are
+ assuming that volatile vector regs are not modified inside
+ __delayLoadHelper2. However, it is known that at least xmm0 and
+ xmm1 are trashed in some versions of Microsoft dlls, and if xmm4 or
+ xmm5 are also used then that would trash the lower bits of ymm4 and
+ ymm5. If it turns out that vector insns with a vex prefix are used
+ then we'll need to save ymm0-5 here but that can't be done without
+ first testing cpuid and xcr0. */
static const char i386_x64_trampoline[] =
- "\tsubq $72, %%rsp\n"
- "\t.seh_stackalloc 72\n"
+ "\tsubq $136, %%rsp\n"
+ "\t.seh_stackalloc 136\n"
"\t.seh_endprologue\n"
- "\tmovq %%rcx, 64(%%rsp)\n"
- "\tmovq %%rdx, 56(%%rsp)\n"
- "\tmovq %%r8, 48(%%rsp)\n"
- "\tmovq %%r9, 40(%%rsp)\n"
- "\tmovq %%rax, %%rdx\n"
- "\tleaq __DELAY_IMPORT_DESCRIPTOR_%s(%%rip), %%rcx\n"
+ "\tmovq %%rcx, 136+8(%%rsp)\n"
+ "\tmovq %%rdx, 136+16(%%rsp)\n"
+ "\tmovq %%r8, 136+24(%%rsp)\n"
+ "\tmovq %%r9, 136+32(%%rsp)\n"
+ "\tmovaps %%xmm0, 32(%%rsp)\n"
+ "\tmovaps %%xmm1, 48(%%rsp)\n"
+ "\tmovaps %%xmm2, 64(%%rsp)\n"
+ "\tmovaps %%xmm3, 80(%%rsp)\n"
+ "\tmovaps %%xmm4, 96(%%rsp)\n"
+ "\tmovaps %%xmm5, 112(%%rsp)\n"
+ "\tmovq %%rax, %%rdx\n"
+ "\tleaq __DELAY_IMPORT_DESCRIPTOR_%s(%%rip), %%rcx\n"
"\tcall __delayLoadHelper2\n"
- "\tmovq 40(%%rsp), %%r9\n"
- "\tmovq 48(%%rsp), %%r8\n"
- "\tmovq 56(%%rsp), %%rdx\n"
- "\tmovq 64(%%rsp), %%rcx\n"
- "\taddq $72, %%rsp\n"
+ "\tmovq 136+8(%%rsp), %%rcx\n"
+ "\tmovq 136+16(%%rsp), %%rdx\n"
+ "\tmovq 136+24(%%rsp), %%r8\n"
+ "\tmovq 136+32(%%rsp), %%r9\n"
+ "\tmovaps 32(%%rsp), %%xmm0\n"
+ "\tmovaps 48(%%rsp), %%xmm1\n"
+ "\tmovaps 64(%%rsp), %%xmm2\n"
+ "\tmovaps 80(%%rsp), %%xmm3\n"
+ "\tmovaps 96(%%rsp), %%xmm4\n"
+ "\tmovaps 112(%%rsp), %%xmm5\n"
+ "\taddq $136, %%rsp\n"
"\tjmp *%%rax\n";
struct mac
reply other threads:[~2023-05-25 7:44 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20230525074445.CA4923858C1F@sourceware.org \
--to=amodra@sourceware.org \
--cc=bfd-cvs@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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).