From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([208.118.235.92]:50506) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1SSsRE-0001D1-48 for qemu-devel@nongnu.org; Fri, 11 May 2012 12:11:29 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1SSsRB-000770-RD for qemu-devel@nongnu.org; Fri, 11 May 2012 12:11:27 -0400 Received: from cantor2.suse.de ([195.135.220.15]:45999 helo=mx2.suse.de) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1SSsRB-00076X-Hg for qemu-devel@nongnu.org; Fri, 11 May 2012 12:11:25 -0400 From: Alexander Graf Date: Fri, 11 May 2012 10:40:10 +0200 Message-Id: <1336725610-8195-1-git-send-email-agraf@suse.de> Subject: [Qemu-devel] [PATCH] linux-user: Fix stale tbs after mmap List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel Developers Cc: Peter Maydell , Riku Voipio If we execute linux-user code that does the following: * A = mmap() * execute code in A * munmap(A) * B = mmap(), but mmap returns the same address as A * execute code in B we end up executing a stale cached tb that contains translated code from A, while we want new code from B. This patch adds a TB flush for mmap'ed regions, before we return them, avoiding the whole issue. It also adds a flush for munmap, so that we don't execute stale TBs instead of getting a segfault. Reported-by: Peter Maydell Signed-off-by: Alexander Graf --- v1 -> v2: - rebase on current git - fix munmap too --- exec-all.h | 2 ++ exec.c | 17 +++++++++++++++++ linux-user/mmap.c | 6 +++++- 3 files changed, 24 insertions(+), 1 deletions(-) diff --git a/exec-all.h b/exec-all.h index c1b7e1f..9bda7f7 100644 --- a/exec-all.h +++ b/exec-all.h @@ -96,6 +96,8 @@ void QEMU_NORETURN cpu_loop_exit(CPUArchState *env1); int page_unprotect(target_ulong address, uintptr_t pc, void *puc); void tb_invalidate_phys_page_range(tb_page_addr_t start, tb_page_addr_t end, int is_cpu_write_access); +void tb_invalidate_phys_range(tb_page_addr_t start, tb_page_addr_t end, + int is_cpu_write_access); #if !defined(CONFIG_USER_ONLY) /* cputlb.c */ void tlb_flush_page(CPUArchState *env, target_ulong addr); diff --git a/exec.c b/exec.c index 0607c9b..a0494c7 100644 --- a/exec.c +++ b/exec.c @@ -1075,6 +1075,23 @@ TranslationBlock *tb_gen_code(CPUArchState *env, return tb; } +/* + * invalidate all TBs which intersect with the target physical pages + * starting in range [start;end[. NOTE: start and end may refer to + * different physical pages. 'is_cpu_write_access' should be true if called + * from a real cpu write access: the virtual CPU will exit the current + * TB if code is modified inside this TB. + */ +void tb_invalidate_phys_range(tb_page_addr_t start, tb_page_addr_t end, + int is_cpu_write_access) +{ + while (start < end) { + tb_invalidate_phys_page_range(start, end, is_cpu_write_access); + start &= TARGET_PAGE_MASK; + start += TARGET_PAGE_SIZE; + } +} + /* invalidate all TBs which intersect with the target physical page starting in range [start;end[. NOTE: start and end must refer to the same physical page. 'is_cpu_write_access' should be true if called diff --git a/linux-user/mmap.c b/linux-user/mmap.c index 7125d1c..d9468fe 100644 --- a/linux-user/mmap.c +++ b/linux-user/mmap.c @@ -573,6 +573,7 @@ abi_long target_mmap(abi_ulong start, abi_ulong len, int prot, page_dump(stdout); printf("\n"); #endif + tb_invalidate_phys_range(start, start + len, 0); mmap_unlock(); return start; fail: @@ -675,8 +676,10 @@ int target_munmap(abi_ulong start, abi_ulong len) } } - if (ret == 0) + if (ret == 0) { page_set_flags(start, start + len, 0); + tb_invalidate_phys_range(start, start + len, 0); + } mmap_unlock(); return ret; } @@ -754,6 +757,7 @@ abi_long target_mremap(abi_ulong old_addr, abi_ulong old_size, page_set_flags(old_addr, old_addr + old_size, 0); page_set_flags(new_addr, new_addr + new_size, prot | PAGE_VALID); } + tb_invalidate_phys_range(new_addr, new_addr + new_size, 0); mmap_unlock(); return new_addr; } -- 1.6.0.2