qemu-stable
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH] target/i386: fix direction of "32-bit MMU" test


From: Michael Tokarev
Subject: Re: [PATCH] target/i386: fix direction of "32-bit MMU" test
Date: Mon, 1 Apr 2024 09:02:51 +0300
User-agent: Mozilla Thunderbird

11.03.2024 10:58, Paolo Bonzini wrote:
The low bit of MMU indices for x86 TCG indicates whether the processor is
in 32-bit mode and therefore linear addresses have to be masked to 32 bits.
However, the index was computed incorrectly, leading to possible conflicts
in the TLB for any address above 4G.

Analyzed-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Fixes: b1661801c18 ("target/i386: Fix physical address truncation", 2024-02-28)
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2206
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Anyone can guess why this rather trivial and obviously correct patch causes 
segfaults
in a few tests in staging-7.2 - when run in tcg mode, namely:

  pxe-test
  migration-test
  boot-serial-test
  bios-tables-test
  vmgenid-test
  cdrom-test

When reverting this single commit from staging-7.2, it all works fine again.

This area in 7.2 is a bit twisty since I picked up previous commit (for which
this one is a fix), b1661801c18 "target/i386: Fix physical address truncation",
before a few others in the same place, especially before 90f641531c782c8
"target/i386: use separate MMU indexes for 32-bit accesses", so that picking
up the other commits has become more problematic due to differences in context.
Also, 7.2 lacks ace0c5fe59 "target/i386: Populate CPUClass.mmu_index" which 
moves
one of the places this commit touches to another file.  Hopefully I haven't
screwed up this place entirely and the prob is just some other missing commit.

I'm re-verifying my cherry-picks once again. Meanwhile, help from someone who
is more familiar with this place is definitely welcome.

BTW, Paolo, it looks like this patch is missing two tags - Reviewed-by Richard
and Tested-by Mark.  It's obviously too late to change that now, but it's
something to check for in the future.

Thanks,

/mjt

---
  target/i386/cpu.h | 2 +-
  target/i386/cpu.c | 2 +-
  2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/target/i386/cpu.h b/target/i386/cpu.h
index 952174bb6f5..6b057380791 100644
--- a/target/i386/cpu.h
+++ b/target/i386/cpu.h
@@ -2334,7 +2334,7 @@ static inline bool is_mmu_index_32(int mmu_index)
static inline int cpu_mmu_index_kernel(CPUX86State *env)
  {
-    int mmu_index_32 = (env->hflags & HF_LMA_MASK) ? 1 : 0;
+    int mmu_index_32 = (env->hflags & HF_LMA_MASK) ? 0 : 1;
      int mmu_index_base =
          !(env->hflags & HF_SMAP_MASK) ? MMU_KNOSMAP64_IDX :
          ((env->hflags & HF_CPL_MASK) < 3 && (env->eflags & AC_MASK)) ? 
MMU_KNOSMAP64_IDX : MMU_KSMAP64_IDX;
diff --git a/target/i386/cpu.c b/target/i386/cpu.c
index 2666ef38089..78524bc6073 100644
--- a/target/i386/cpu.c
+++ b/target/i386/cpu.c
@@ -7735,7 +7735,7 @@ static bool x86_cpu_has_work(CPUState *cs)
  static int x86_cpu_mmu_index(CPUState *cs, bool ifetch)
  {
      CPUX86State *env = cpu_env(cs);
-    int mmu_index_32 = (env->hflags & HF_CS64_MASK) ? 1 : 0;
+    int mmu_index_32 = (env->hflags & HF_CS64_MASK) ? 0 : 1;
      int mmu_index_base =
          (env->hflags & HF_CPL_MASK) == 3 ? MMU_USER64_IDX :
          !(env->hflags & HF_SMAP_MASK) ? MMU_KNOSMAP64_IDX :




reply via email to

[Prev in Thread] Current Thread [Next in Thread]