qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Ziqiao Kong <ziqiaokong@gmail.com>
To: qemu-devel@nongnu.org
Cc: Ziqiao Kong <ziqiaokong@gmail.com>,
	qemu-trivial@nongnu.org,
	Richard Henderson <richard.henderson@linaro.org>
Subject: [PATCH 0/1] Fix tcg assert when decoding "lock cmp"
Date: Thu, 15 Feb 2024 17:29:08 +0800	[thread overview]
Message-ID: <20240215092908.569947-1-ziqiaokong@gmail.com> (raw)

(Resending this series since it looks like being ignored by the list)

Hello QEMU Developers! I'm from the downstream project Unicorn
Engine (https://github.com/unicorn-engine/unicorn). I identified a simple
bug that triggers a tcg assertion in QEMU. Although I used a usermode
usecase to illustrate the bug, it shall also affect system emulation.

Assume a binary looks like:

```
08049845 <main>:
 8049845:	55                   	push   %ebp
 8049846:	89 e5                	mov    %esp,%ebp
 8049848:	e8 0e 00 00 00       	call   804985b <__x86.get_pc_thunk.ax>
 804984d:	90                   	nop
 804984e:	90                   	nop
 804984f:	90                   	nop
 8049850:	90                   	nop
 8049851:	90                   	nop
 8049852:	f0 38 b8 00 00 00 00 	lock cmp %bh,0x0(%eax)
 8049859:	5d                   	pop    %ebp
 804985a:	c3                   	ret
```

Executing this with

```
./i386-linux-user/qemu-i386 -d op,op_opt ./tt 2>&1 | tail -100
```

gives:

```
OP:
...
...
 ---- 0000000008049852 0000000000000000
 mov_i32 loc2,eax
 extract_i32 loc1,ebx,$0x8,$0x8
 mov_i32 cc_src,loc1
 mov_i32 loc8,loc0 // loc0 undefined!
...
...
OP after optimization and liveness analysis:
...
...
 ---- 0000000008049852 0000000000000000
 extract_i32 tmp1,ebx,$0x8,$0x8           dead: 1  pref=0xffff
 mov_i32 cc_src,tmp1                      sync: 0  dead: 1  pref=0xffff
 sub_i32 cc_dst,tmp0,cc_src               sync: 0  dead: 0 1 2  pref=0xffff // so does tmp0!
...
...
ERROR:../tcg/tcg.c:4446:temp_load: code should not be reached
Bail out! ERROR:../tcg/tcg.c:4446:temp_load: code should not be reached
```

This results an assertion error because when decoding "lock cmp %bh,0x0(%eax)",
s->T0 is not properly initialized in gen_op. However, the root cause is that
this instruction is illegal refering to the Intel manual. Therefore, I add
and extra check to generate corresponding illegal opcode exception.

Signed-off-by: Ziqiao Kong <ziqiaokong@gmail.com>

Ziqiao Kong (1):
  Generate an illegal opcode exception on cmp instructions with lock
    prefix

 target/i386/tcg/translate.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

-- 
2.40.1



             reply	other threads:[~2024-02-15 13:28 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-02-15  9:29 Ziqiao Kong [this message]
2024-02-15  9:33 ` [PATCH 1/1] target/i386: Generate an illegal opcode exception on cmp instructions with lock prefix Ziqiao Kong
  -- strict thread matches above, loose matches on Subject: below --
2024-02-15  9:50 [PATCH 0/1] Fix tcg assert when decoding "lock cmp" Ziqiao Kong
2024-02-15  7:56 Ziqiao Kong

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=20240215092908.569947-1-ziqiaokong@gmail.com \
    --to=ziqiaokong@gmail.com \
    --cc=qemu-devel@nongnu.org \
    --cc=qemu-trivial@nongnu.org \
    --cc=richard.henderson@linaro.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 NNTP newsgroup(s).