diff options
| author |   Jan Beulich <jbeulich@suse.com> | 2019-11-11 15:32:12 +0100 |
|---|---|---|
| committer |   Thomas Gleixner <tglx@linutronix.de> | 2019-11-19 21:58:28 +0100 |
| commit | 29b810f5a5ec127d3143770098e05981baa3eb77 (patch) | |
| tree | da09a8c8e9d7ad026bcb3d99222464a91aa16dd0 /arch/x86/entry | |
| parent | x86/stackframe/32: Repair 32-bit Xen PV (diff) | |
| download | linux-dev-29b810f5a5ec127d3143770098e05981baa3eb77.tar.xz linux-dev-29b810f5a5ec127d3143770098e05981baa3eb77.zip | |
x86/xen/32: Make xen_iret_crit_fixup() independent of frame layout
Now that SS:ESP always get saved by SAVE_ALL, this also needs to be
accounted for in xen_iret_crit_fixup(). Otherwise the old_ax value gets
interpreted as EFLAGS, and hence VM86 mode appears to be active all the
time, leading to random "vm86_32: no user_vm86: BAD" log messages alongside
processes randomly crashing.
Since following the previous model (sitting after SAVE_ALL) would further
complicate the code _and_ retain the dependency of xen_iret_crit_fixup() on
frame manipulations done by entry_32.S, switch things around and do the
adjustment ahead of SAVE_ALL.
Fixes: 3c88c692c287 ("x86/stackframe/32: Provide consistent pt_regs")
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Juergen Gross <jgross@suse.com>
Cc: Stable Team <stable@vger.kernel.org>
Link: https://lkml.kernel.org/r/32d8713d-25a7-84ab-b74b-aa3e88abce6b@suse.com
Diffstat (limited to 'arch/x86/entry')
| -rw-r--r-- | arch/x86/entry/entry_32.S | 22 |
1 files changed, 9 insertions, 13 deletions
diff --git a/arch/x86/entry/entry_32.S b/arch/x86/entry/entry_32.S index 3f847d843e47..019dbac6fe9d 100644 --- a/arch/x86/entry/entry_32.S +++ b/arch/x86/entry/entry_32.S @@ -1341,11 +1341,6 @@ END(spurious_interrupt_bug) #ifdef CONFIG_XEN_PV ENTRY(xen_hypervisor_callback) - pushl $-1 /* orig_ax = -1 => not a system call */ - SAVE_ALL - ENCODE_FRAME_POINTER - TRACE_IRQS_OFF - /* * Check to see if we got the event in the critical * region in xen_iret_direct, after we've reenabled @@ -1353,16 +1348,17 @@ ENTRY(xen_hypervisor_callback) * iret instruction's behaviour where it delivers a * pending interrupt when enabling interrupts: */ - movl PT_EIP(%esp), %eax - cmpl $xen_iret_start_crit, %eax + cmpl $xen_iret_start_crit, (%esp) jb 1f - cmpl $xen_iret_end_crit, %eax + cmpl $xen_iret_end_crit, (%esp) jae 1f - - jmp xen_iret_crit_fixup - -ENTRY(xen_do_upcall) -1: mov %esp, %eax + call xen_iret_crit_fixup +1: + pushl $-1 /* orig_ax = -1 => not a system call */ + SAVE_ALL + ENCODE_FRAME_POINTER + TRACE_IRQS_OFF + mov %esp, %eax call xen_evtchn_do_upcall #ifndef CONFIG_PREEMPTION call xen_maybe_preempt_hcall |