From: Paolo Bonzini <pbonzini@redhat.com>
To: qemu-devel@nongnu.org
Cc: richard.henderson@linaro.org
Subject: [Qemu-devel] [PATCH 2/2] target/i386: trap on instructions longer than >15 bytes
Date: Thu, 12 Oct 2017 16:35:48 +0200 [thread overview]
Message-ID: <20171012143548.18581-3-pbonzini@redhat.com> (raw)
In-Reply-To: <20171012143548.18581-1-pbonzini@redhat.com>
Besides being more correct, arbitrarily long instruction allow the
generation of a translation block that spans three pages. This
confuses the generator and even allows ring 3 code to poison the
translation block cache and inject code into other processes that are
in guest ring 3.
This is an improved (and more invasive) fix for the bug fixed in commit
30663fd ("tcg/i386: Check the size of instruction being translated",
2017-03-24). In addition to being more precise (and generating the
right exception, which is #GP rather than #UD), it distinguishes better
between page faults and too long instructions, as shown by this test case:
#include <sys/mman.h>
#include <string.h>
#include <stdio.h>
int main()
{
char *x = mmap(NULL, 8192, PROT_READ|PROT_WRITE|PROT_EXEC,
MAP_PRIVATE|MAP_ANON, -1, 0);
memset(x, 0x66, 4096);
x[4096] = 0x90;
x[4097] = 0xc3;
char *i = x + 4096 - 15;
mprotect(x + 4096, 4096, PROT_READ|PROT_WRITE);
((void(*)(void)) i) ();
}
... which produces a #GP without the mprotect, and a #PF with it.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
target/i386/translate.c | 29 ++++++++++++++++++++++-------
1 file changed, 22 insertions(+), 7 deletions(-)
diff --git a/target/i386/translate.c b/target/i386/translate.c
index 4a938c21a0..5f24a2de3c 100644
--- a/target/i386/translate.c
+++ b/target/i386/translate.c
@@ -136,6 +136,7 @@ typedef struct DisasContext {
int cpuid_ext3_features;
int cpuid_7_0_ebx_features;
int cpuid_xsave_features;
+ sigjmp_buf jmpbuf;
} DisasContext;
static void gen_eob(DisasContext *s);
@@ -1863,11 +1864,27 @@ static void gen_shifti(DisasContext *s1, int op, TCGMemOp ot, int d, int c)
}
}
+#define X86_MAX_INSN_LENGTH 15
+
static uint64_t advance_pc(CPUX86State *env, DisasContext *s, int num_bytes)
{
uint64_t pc = s->pc;
s->pc += num_bytes;
+ if (unlikely(s->pc - s->pc_start > X86_MAX_INSN_LENGTH)) {
+ /* If the instruction's 16th byte is on a different page than the 1st, a
+ * page fault on the second page wins over the general protection fault
+ * caused by the instruction being too long.
+ * This can happen even if the operand is only one byte long!
+ */
+ if (((s->pc - 1) ^ (pc - 1)) & TARGET_PAGE_MASK) {
+ volatile uint8_t unused =
+ cpu_ldub_code(env, (s->pc - 1) & TARGET_PAGE_MASK);
+ (void) unused;
+ }
+ siglongjmp(s->jmpbuf, 1);
+ }
+
return pc;
}
@@ -4463,14 +4480,12 @@ static target_ulong disas_insn(DisasContext *s, CPUState *cpu)
s->rip_offset = 0; /* for relative ip address */
s->vex_l = 0;
s->vex_v = 0;
- next_byte:
- /* x86 has an upper limit of 15 bytes for an instruction. Since we
- * do not want to decode and generate IR for an illegal
- * instruction, the following check limits the instruction size to
- * 25 bytes: 14 prefix + 1 opc + 6 (modrm+sib+ofs) + 4 imm */
- if (s->pc - pc_start > 14) {
- goto illegal_op;
+ if (sigsetjmp(s->jmpbuf, 0) != 0) {
+ gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base);
+ return s->pc;
}
+
+ next_byte:
b = x86_ldub_code(env, s);
/* Collect prefixes. */
switch (b) {
--
2.14.2
next prev parent reply other threads:[~2017-10-12 14:35 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-10-12 14:35 [Qemu-devel] [PATCH 0/2] target/i386: trap on instructions longer than >15 bytes Paolo Bonzini
2017-10-12 14:35 ` [Qemu-devel] [PATCH 1/2] target/i386: introduce x86_ld*_code Paolo Bonzini
2017-10-12 16:30 ` Richard Henderson
2017-10-12 14:35 ` Paolo Bonzini [this message]
2017-10-12 19:30 ` [Qemu-devel] [PATCH 2/2] target/i386: trap on instructions longer than >15 bytes Richard Henderson
2017-10-12 21:43 ` Paolo Bonzini
2017-10-13 18:12 ` Richard Henderson
2017-10-12 16:55 ` [Qemu-devel] [PATCH 0/2] " no-reply
2017-10-12 19:07 ` no-reply
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=20171012143548.18581-3-pbonzini@redhat.com \
--to=pbonzini@redhat.com \
--cc=qemu-devel@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).