|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Re: [PATCH v2] x86emul: Zero-extend unsigned 8-bit immediates when decoding
On 16.11.2026 02:27, Stephen Cheng wrote:
> The decoder sign-extends all 8-bit immediates. This is right for e.g.
> Jcc rel8 and Grp1 (83), but some 8-bit immediates are unsigned: the
> vector of INT, the port of IN/OUT, the base of AAM/AAD, the count of
> the Grp2 shifts (C0, C1), and the 8-bit immediates in the 0F, 0F3A and
> XOP08 maps. For these, s->imm1 holds a wrong value for 0x80 and above.
> x86_insn_immediate() can return this value too.
>
> The worst case is INT $0x80 to INT $0xff, which puts a negative vector
> into the event (e.g. -128 for 0x80). hvm_inject_event() then fails its
> "vector == event->vector" assertion, and a debug Xen panics. An HVM
> guest can trigger this with the forced emulation prefix (hvm_fep).
>
> Zero-extend these immediates in the decoder, and drop the (uint8_t)
> casts at the IN/OUT and AAM/AAD use sites, which are not needed any
> more.
>
> Fixes: 653cae725cea ("x86/32on64: use generic instruction decoding for call
> gate emulation")
> Fixes: 52053d5fc2f9 ("x86/emul: Drop swint_emulate infrastructure")
> Assisted-by: Claude:claude-opus-5-5
> Signed-off-by: Stephen Cheng <stephen.cheng@xxxxxxxxxx>
Looks largely good to me now, just a few minor things:
> --- a/xen/arch/x86/x86_emulate/decode.c
> +++ b/xen/arch/x86/x86_emulate/decode.c
> @@ -650,10 +650,19 @@ decode_onebyte(struct x86_emulate_state *s,
> case 0xc4: /* les */
> case 0xc5: /* lds */
> case 0xce: /* into */
> + case 0xd6: /* salc / udb */
> + s->not_64bit = true;
> + break;
> +
> case 0xd4: /* aam */
> case 0xd5: /* aad */
> - case 0xd6: /* salc / udb */
> s->not_64bit = true;
> + fallthrough;
> + case 0xc0: case 0xc1: /* Grp2 */
> + case 0xcd: /* int imm8 */
> + case 0xe4 ... 0xe7: /* in/out imm8 */
> + /* The 8-bit immediate is unsigned. */
> + s->imm1 = (uint8_t)s->imm1;
> break;
This now separate case block needs to move down. All such switch() statements
aim at having their case blocks sorted by lowest-numbered case each. There
may be anomalies to that rule, but we want to make sure to reduce their set,
not increase it.
> @@ -1796,6 +1805,8 @@ int x86emul_decode(struct x86_emulate_state *s,
> break;
>
> case ext_0f:
> + if ( (d & SrcMask) == SrcImmByte )
> + s->imm1 = (uint8_t)s->imm1;
> rc = decode_twobyte(s, ctxt, ops);
> break;
>
> @@ -1804,6 +1815,7 @@ int x86emul_decode(struct x86_emulate_state *s,
> break;
>
> case ext_0f3a:
> + s->imm1 = (uint8_t)s->imm1;
> d = ext0f3a_table[b].to_mem ? DstMem | SrcReg : DstReg | SrcMem;
> if ( ext0f3a_table[b].two_op )
> d |= TwoOp;
> @@ -1814,6 +1826,7 @@ int x86emul_decode(struct x86_emulate_state *s,
> break;
>
> case ext_8f08:
> + s->imm1 = (uint8_t)s->imm1;
> d = DstReg | SrcMem;
> if ( ext8f08_table[b].two_op )
> d |= TwoOp;
All of these three may benefit from having a brief comment added, e.g.
"All immediates in this encoding space are unsigned ones."
Also, just to mention - in the APX series, where I've done the respective
adjustments already, I've deliberately avoided the use of casts. "&= 0xff"
is shorter, and avoiding casts is generally a worthwhile goal imo.
> --- a/xen/arch/x86/x86_emulate/x86_emulate.c
> +++ b/xen/arch/x86/x86_emulate/x86_emulate.c
> @@ -2505,7 +2505,7 @@ x86_emulate(
>
> case 0xd4: /* aam */
> case 0xd5: /* aad */
> - n = (uint8_t)src.val;
> + n = src.val;
> if ( b & 0x01 )
> _regs.ax = (uint8_t)(_regs.al + (_regs.ah * n));
> else
> @@ -2561,7 +2561,7 @@ x86_emulate(
> case 0xed: /* in %dx,%eax */
> case 0xee: /* out %al,%dx */
> case 0xef: /* out %eax,%dx */ {
> - unsigned int port = ((b < 0xe8) ? (uint8_t)src.val : _regs.dx);
> + unsigned int port = ((b < 0xe8) ? src.val : _regs.dx);
>
> op_bytes = !(b & 1) ? 1 : (op_bytes == 8) ? 4 : op_bytes;
> if ( (rc = ioport_access_check(port, op_bytes, ctxt, ops)) != 0 )
How about an ASSERT() each, to make sure the handling doesn't break again?
Jan
|
![]() |
Lists.xenproject.org is hosted with RackSpace, monitoring our |