On Wed, Jul 10, 2024 at 11:50:50PM +0200, Uros Bizjak wrote:
On 10. 07. 24 21:06, Pawan Gupta wrote:
Robert Gill reported below #GP when dosemu software was executing vm86() system call:
general protection fault: 0000 [#1] PREEMPT SMP CPU: 4 PID: 4610 Comm: dosemu.bin Not tainted 6.6.21-gentoo-x86 #1 Hardware name: Dell Inc. PowerEdge 1950/0H723K, BIOS 2.7.0 10/30/2010 EIP: restore_all_switch_stack+0xbe/0xcf EAX: 00000000 EBX: 00000000 ECX: 00000000 EDX: 00000000 ESI: 00000000 EDI: 00000000 EBP: 00000000 ESP: ff8affdc DS: 0000 ES: 0000 FS: 0000 GS: 0033 SS: 0068 EFLAGS: 00010046 CR0: 80050033 CR2: 00c2101c CR3: 04b6d000 CR4: 000406d0 Call Trace: show_regs+0x70/0x78 die_addr+0x29/0x70 exc_general_protection+0x13c/0x348 exc_bounds+0x98/0x98 handle_exception+0x14d/0x14d exc_bounds+0x98/0x98 restore_all_switch_stack+0xbe/0xcf exc_bounds+0x98/0x98 restore_all_switch_stack+0xbe/0xcf
This only happens when VERW based mitigations like MDS/RFDS are enabled. This is because segment registers with an arbitrary user value can result in #GP when executing VERW. Intel SDM vol. 2C documents the following behavior for VERW instruction:
#GP(0) - If a memory operand effective address is outside the CS, DS, ES, FS, or GS segment limit.
CLEAR_CPU_BUFFERS macro executes VERW instruction before returning to user space. Replace CLEAR_CPU_BUFFERS with a safer version that uses %ss to refer VERW operand mds_verw_sel. This ensures VERW will not #GP for an arbitrary user %ds. Also, in NMI return path, move VERW to after RESTORE_ALL_NMI that touches GPRs.
For clarity, below are the locations where the new CLEAR_CPU_BUFFERS_SAFE version is being used:
- entry_INT80_32(), entry_SYSENTER_32() and interrupts (via handle_exception_return) do:
restore_all_switch_stack: [...] mov %esi,%esi verw %ss:0xc0fc92c0 <------------- iret
Opportunistic SYSEXIT:
[...] verw %ss:0xc0fc92c0 <------------- btrl $0x9,(%esp) popf pop %eax sti sysexit
nmi_return and nmi_from_espfix: mov %esi,%esi verw %ss:0xc0fc92c0 <------------- jmp .Lirq_return
Fixes: a0e2dab44d22 ("x86/entry_32: Add VERW just before userspace transition") Cc: stable@vger.kernel.org # 5.10+ Reported-by: Robert Gill rtgill82@gmail.com Closes: https://bugzilla.kernel.org/show_bug.cgi?id=218707 Closes: https://lore.kernel.org/all/8c77ccfd-d561-45a1-8ed5-6b75212c7a58@leemhuis.in... Suggested-by: Dave Hansen dave.hansen@linux.intel.com Suggested-by: Brian Gerst brgerst@gmail.com # Use %ss Signed-off-by: Pawan Gupta pawan.kumar.gupta@linux.intel.com
v4:
- Further simplify the patch by using %ss for all VERW calls in 32-bit mode (Brian).
- In NMI exit path move VERW after RESTORE_ALL_NMI that touches GPRs (Dave).
v3: https://lore.kernel.org/r/20240701-fix-dosemu-vm86-v3-1-b1969532c75a@linux.i...
- Simplify CLEAR_CPU_BUFFERS_SAFE by using %ss instead of %ds (Brian).
- Do verw before popf in SYSEXIT path (Jari).
v2: https://lore.kernel.org/r/20240627-fix-dosemu-vm86-v2-1-d5579f698e77@linux.i...
- Safe guard against any other system calls like vm86() that might change %ds (Dave).
v1: https://lore.kernel.org/r/20240426-fix-dosemu-vm86-v1-1-88c826a3f378@linux.i...
arch/x86/entry/entry_32.S | 18 +++++++++++++++--- 1 file changed, 15 insertions(+), 3 deletions(-)
diff --git a/arch/x86/entry/entry_32.S b/arch/x86/entry/entry_32.S index d3a814efbff6..d54f6002e5a0 100644 --- a/arch/x86/entry/entry_32.S +++ b/arch/x86/entry/entry_32.S @@ -253,6 +253,16 @@ .Lend_@: .endm +/*
- Safer version of CLEAR_CPU_BUFFERS that uses %ss to reference VERW operand
- mds_verw_sel. This ensures VERW will not #GP for an arbitrary user %ds.
- */
+.macro CLEAR_CPU_BUFFERS_SAFE
- ALTERNATIVE "jmp .Lskip_verw@", "", X86_FEATURE_CLEAR_CPU_BUF
- verw %ss:_ASM_RIP(mds_verw_sel)
+.Lskip_verw@: +.endm
Why not simply:
.macro CLEAR_CPU_BUFFERS_SAFE ALTERNATIVE "", __stringify(verw %ss:_ASM_RIP(mds_verw_sel)), X86_FEATURE_CLEAR_CPU_BUF .endm
We can do it this way as well. But, there are stable kernels that don't support relocations in ALTERNATIVEs. The way it is done in current patch can be backported without worrying about which kernels support relocations.