diff mbox

Fix for self-modifying writes across page boundaries.

Message ID 20160706182652.16190-1-samuel.damashek@invincea.com (mailing list archive)
State New, archived
Headers show

Commit Message

Samuel Damashek July 6, 2016, 6:26 p.m. UTC
As it currently stands, QEMU does not properly handle self-modifying code
when the write is unaligned and crosses a page boundary. The procedure
for handling a write to the current translation block is to write-protect
the current translation block, catch the write, split up the translation
block into the current instruction (which remains write-protected so that
the current instruction is not modified) and the remaining instructions
in the translation block, and then restore the CPU state to before the
write occurred so the write will be retried and successfully executed.
However, since unaligned writes across pages are split into one-byte
writes for simplicity, writes to the second page (which is not the
current TB) may succeed before a write to the current TB is attempted,
and since these writes are not invalidated before resuming state after
splitting the TB, these writes will be performed a second time, thus
corrupting the second page. Credit goes to Patrick Hulin for
discovering this.

In recent 64-bit versions of Windows running in emulated mode, this
results in either being very unstable (a BSOD after a couple minutes of
uptime), or being entirely unable to boot. Windows performs one or more
8-byte unaligned self-modifying writes (xors) which intersect the end
of the current TB and the beginning of the next TB, which runs into the
aforementioned issue. This commit fixes that issue by making the
unaligned write loop perform the writes in forwards order, instead of
reverse order. This way, QEMU immediately tries to write to the current
TB, and splits the TB before any write to the second page is executed.
The write then proceeds as intended. With this patch applied, I am able
to boot and use Windows 7 64-bit and Windows 10 64-bit in QEMU without
KVM.

Per Richard Henderson's input, this patch also ensures the second page
is in the TLB before executing the write loop, to ensure the second
page is mapped.

The VICTIM_TLB_HIT macro was also updated to accept a second argument,
the name of the addr variable being used, so that I could add an
additional addr variable (addr_page2) instead of temporarily setting addr
to the second page's address.

The original discussion of the issue is located at
http://lists.nongnu.org/archive/html/qemu-devel/2014-08/msg02161.html.

Signed-off-by: Samuel Damashek <samuel.damashek@invincea.com>
---
 softmmu_template.h | 61 ++++++++++++++++++++++++++++++++++++++++--------------
 1 file changed, 46 insertions(+), 15 deletions(-)
diff mbox

Patch

diff --git a/softmmu_template.h b/softmmu_template.h
index 208f808..0d173a0 100644
--- a/softmmu_template.h
+++ b/softmmu_template.h
@@ -117,7 +117,7 @@ 
 #endif
 
 /* macro to check the victim tlb */
-#define VICTIM_TLB_HIT(ty)                                                    \
+#define VICTIM_TLB_HIT(ty, addr_nm)                                           \
 ({                                                                            \
     /* we are about to do a page table walk. our last hope is the             \
      * victim tlb. try to refill from the victim tlb before walking the       \
@@ -126,7 +126,8 @@ 
     CPUIOTLBEntry tmpiotlb;                                                   \
     CPUTLBEntry tmptlb;                                                       \
     for (vidx = CPU_VTLB_SIZE-1; vidx >= 0; --vidx) {                         \
-        if (env->tlb_v_table[mmu_idx][vidx].ty == (addr & TARGET_PAGE_MASK)) {\
+        if (env->tlb_v_table[mmu_idx][vidx].ty                                \
+            == (addr_nm & TARGET_PAGE_MASK)) {                                \
             /* found entry in victim tlb, swap tlb and iotlb */               \
             tmptlb = env->tlb_table[mmu_idx][index];                          \
             env->tlb_table[mmu_idx][index] = env->tlb_v_table[mmu_idx][vidx]; \
@@ -185,7 +186,7 @@  WORD_TYPE helper_le_ld_name(CPUArchState *env, target_ulong addr,
             cpu_unaligned_access(ENV_GET_CPU(env), addr, READ_ACCESS_TYPE,
                                  mmu_idx, retaddr);
         }
-        if (!VICTIM_TLB_HIT(ADDR_READ)) {
+        if (!VICTIM_TLB_HIT(ADDR_READ, addr)) {
             tlb_fill(ENV_GET_CPU(env), addr, READ_ACCESS_TYPE,
                      mmu_idx, retaddr);
         }
@@ -269,7 +270,7 @@  WORD_TYPE helper_be_ld_name(CPUArchState *env, target_ulong addr,
             cpu_unaligned_access(ENV_GET_CPU(env), addr, READ_ACCESS_TYPE,
                                  mmu_idx, retaddr);
         }
-        if (!VICTIM_TLB_HIT(ADDR_READ)) {
+        if (!VICTIM_TLB_HIT(ADDR_READ, addr)) {
             tlb_fill(ENV_GET_CPU(env), addr, READ_ACCESS_TYPE,
                      mmu_idx, retaddr);
         }
@@ -389,7 +390,7 @@  void helper_le_st_name(CPUArchState *env, target_ulong addr, DATA_TYPE val,
             cpu_unaligned_access(ENV_GET_CPU(env), addr, MMU_DATA_STORE,
                                  mmu_idx, retaddr);
         }
-        if (!VICTIM_TLB_HIT(addr_write)) {
+        if (!VICTIM_TLB_HIT(addr_write, addr)) {
             tlb_fill(ENV_GET_CPU(env), addr, MMU_DATA_STORE, mmu_idx, retaddr);
         }
         tlb_addr = env->tlb_table[mmu_idx][index].addr_write;
@@ -414,16 +415,31 @@  void helper_le_st_name(CPUArchState *env, target_ulong addr, DATA_TYPE val,
     if (DATA_SIZE > 1
         && unlikely((addr & ~TARGET_PAGE_MASK) + DATA_SIZE - 1
                      >= TARGET_PAGE_SIZE)) {
-        int i;
+        int i, index_page2;
+        target_ulong addr_page2, tlb_addr_page2;
     do_unaligned_access:
         if ((get_memop(oi) & MO_AMASK) == MO_ALIGN) {
             cpu_unaligned_access(ENV_GET_CPU(env), addr, MMU_DATA_STORE,
                                  mmu_idx, retaddr);
         }
+
+        /* Ensure the second page is in the TLB. Note that the first page
+         * is already guaranteed to be filled into the TLB. */
+        addr_page2 = (addr + DATA_SIZE) & TARGET_PAGE_MASK;
+        index_page2 = (addr_page2 >> TARGET_PAGE_BITS) & (CPU_TLB_SIZE - 1);
+        tlb_addr_page2 = env->tlb_table[mmu_idx][index_page2].addr_write;
+
+        if (((addr_page2 & TARGET_PAGE_MASK)
+            != (tlb_addr_page2 & (TARGET_PAGE_MASK | TLB_INVALID_MASK)))
+            && !VICTIM_TLB_HIT(addr_write, addr_page2)) {
+            tlb_fill(ENV_GET_CPU(env), addr_page2, MMU_DATA_STORE,
+                             mmu_idx, retaddr);
+        }
+
         /* XXX: not efficient, but simple */
-        /* Note: relies on the fact that tlb_fill() does not remove the
-         * previous page from the TLB cache.  */
-        for (i = DATA_SIZE - 1; i >= 0; i--) {
+        /* This loop must go in the forward direction to avoid issues with
+         * self-modifying code in Windows 64-bit. */
+        for (i = 0; i < DATA_SIZE; i++) {
             /* Little-endian extract.  */
             uint8_t val8 = val >> (i * 8);
             /* Note the adjustment at the beginning of the function.
@@ -469,7 +485,7 @@  void helper_be_st_name(CPUArchState *env, target_ulong addr, DATA_TYPE val,
             cpu_unaligned_access(ENV_GET_CPU(env), addr, MMU_DATA_STORE,
                                  mmu_idx, retaddr);
         }
-        if (!VICTIM_TLB_HIT(addr_write)) {
+        if (!VICTIM_TLB_HIT(addr_write, addr)) {
             tlb_fill(ENV_GET_CPU(env), addr, MMU_DATA_STORE, mmu_idx, retaddr);
         }
         tlb_addr = env->tlb_table[mmu_idx][index].addr_write;
@@ -494,16 +510,31 @@  void helper_be_st_name(CPUArchState *env, target_ulong addr, DATA_TYPE val,
     if (DATA_SIZE > 1
         && unlikely((addr & ~TARGET_PAGE_MASK) + DATA_SIZE - 1
                      >= TARGET_PAGE_SIZE)) {
-        int i;
+        int i, index_page2;
+        target_ulong addr_page2, tlb_addr_page2;
     do_unaligned_access:
         if ((get_memop(oi) & MO_AMASK) == MO_ALIGN) {
             cpu_unaligned_access(ENV_GET_CPU(env), addr, MMU_DATA_STORE,
                                  mmu_idx, retaddr);
         }
+
+        /* Ensure the second page is in the TLB. Note that the first page
+         * is already guaranteed to be filled into the TLB. */
+        addr_page2 = (addr + DATA_SIZE) & TARGET_PAGE_MASK;
+        index_page2 = (addr_page2 >> TARGET_PAGE_BITS) & (CPU_TLB_SIZE - 1);
+        tlb_addr_page2 = env->tlb_table[mmu_idx][index_page2].addr_write;
+
+        if (((addr_page2 & TARGET_PAGE_MASK)
+            != (tlb_addr_page2 & (TARGET_PAGE_MASK | TLB_INVALID_MASK)))
+            && !VICTIM_TLB_HIT(addr_write, addr_page2)) {
+            tlb_fill(ENV_GET_CPU(env), addr_page2, MMU_DATA_STORE,
+                             mmu_idx, retaddr);
+        }
+
         /* XXX: not efficient, but simple */
-        /* Note: relies on the fact that tlb_fill() does not remove the
-         * previous page from the TLB cache.  */
-        for (i = DATA_SIZE - 1; i >= 0; i--) {
+        /* This loop must go in the forward direction to avoid issues with
+         * self-modifying code in Windows 64-bit. */
+        for (i = 0; i < DATA_SIZE; i++) {
             /* Big-endian extract.  */
             uint8_t val8 = val >> (((DATA_SIZE - 1) * 8) - (i * 8));
             /* Note the adjustment at the beginning of the function.
@@ -542,7 +573,7 @@  void probe_write(CPUArchState *env, target_ulong addr, int mmu_idx,
     if ((addr & TARGET_PAGE_MASK)
         != (tlb_addr & (TARGET_PAGE_MASK | TLB_INVALID_MASK))) {
         /* TLB entry is for a different page */
-        if (!VICTIM_TLB_HIT(addr_write)) {
+        if (!VICTIM_TLB_HIT(addr_write, addr)) {
             tlb_fill(ENV_GET_CPU(env), addr, MMU_DATA_STORE, mmu_idx, retaddr);
         }
     }