From: Ilya Leoshkevich <iii@linux.ibm.com>
To: "Laurent Vivier" <laurent@vivier.eu>,
"Eduardo Habkost" <eduardo@habkost.net>,
"Marcel Apfelbaum" <marcel.apfelbaum@gmail.com>,
"Philippe Mathieu-Daudé" <f4bug@amsat.org>,
"Yanan Wang" <wangyanan55@huawei.com>,
"Richard Henderson" <richard.henderson@linaro.org>,
"Paolo Bonzini" <pbonzini@redhat.com>,
"David Hildenbrand" <david@redhat.com>
Cc: qemu-devel@nongnu.org, qemu-s390x@nongnu.org,
Christian Borntraeger <borntraeger@linux.ibm.com>,
Ilya Leoshkevich <iii@linux.ibm.com>
Subject: [PATCH 1/2] linux-user: Fix siginfo_t contents when jumping to non-readable pages
Date: Thu, 4 Aug 2022 20:23:58 +0200 [thread overview]
Message-ID: <20220804182359.830058-2-iii@linux.ibm.com> (raw)
In-Reply-To: <20220804182359.830058-1-iii@linux.ibm.com>
When the first instruction of a translation block is located in a
non-readable page, qemu-user fills siginfo_t correctly. For the other
instructions the result is as if it were the first instruction, which
is not correct.
The reason is that the current logic expects translate_insn() hook to
stop at the page boundary. This way only the first instruction can
cause a SEGV. However, this is quite difficult to properly implement
when the problematic instruction crosses a page boundary, and indeed
the actual implementations do not do this. Note that this can also
break self-modifying code detection when only bytes on the second page
are modified, but this is outside of the scope of this patch.
Instead of chaning all the translators, do a much simpler thing: when
such a situation is detected, start from scratch and stop right before
the problematic instruction.
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
---
accel/tcg/translate-all.c | 16 +++++++++++-----
accel/tcg/translator.c | 25 +++++++++++++++++++++++++
include/hw/core/cpu.h | 2 ++
linux-user/signal.c | 5 +++++
4 files changed, 43 insertions(+), 5 deletions(-)
diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
index ef62a199c7..b4766f4661 100644
--- a/accel/tcg/translate-all.c
+++ b/accel/tcg/translate-all.c
@@ -2295,12 +2295,18 @@ void page_set_flags(target_ulong start, target_ulong end, int flags)
len != 0;
len -= TARGET_PAGE_SIZE, addr += TARGET_PAGE_SIZE) {
PageDesc *p = page_find_alloc(addr >> TARGET_PAGE_BITS, 1);
+ bool invalidate;
- /* If the write protection bit is set, then we invalidate
- the code inside. */
- if (!(p->flags & PAGE_WRITE) &&
- (flags & PAGE_WRITE) &&
- p->first_tb) {
+ /*
+ * If the write protection bit is set, then we invalidate the code
+ * inside. For qemu-user, we need to do this when PAGE_READ is cleared
+ * as well, in order to force a SEGV when trying to run this code.
+ */
+ invalidate = !(p->flags & PAGE_WRITE) && (flags & PAGE_WRITE);
+#ifdef CONFIG_USER_ONLY
+ invalidate |= (p->flags & PAGE_READ) && !(flags & PAGE_READ);
+#endif
+ if (invalidate && p->first_tb) {
tb_invalidate_phys_page(addr, 0);
}
if (reset_target_data) {
diff --git a/accel/tcg/translator.c b/accel/tcg/translator.c
index fe7af9b943..e444c17515 100644
--- a/accel/tcg/translator.c
+++ b/accel/tcg/translator.c
@@ -57,6 +57,18 @@ void translator_loop(const TranslatorOps *ops, DisasContextBase *db,
uint32_t cflags = tb_cflags(tb);
bool plugin_enabled;
+ /*
+ * In case translate_insn hook touched an unreadable page, redo the
+ * translation until the problematic instruction. We cannot just throw
+ * away the trailing ops, because the hook could have changed DisasContext.
+ */
+ tcg_debug_assert(!cpu->translator_jmp);
+ if (sigsetjmp(cpu->translator_jmp_env, 1) != 0) {
+ cpu->translator_jmp = false;
+ tcg_remove_ops_after(NULL);
+ max_insns = db->num_insns - 1;
+ }
+
/* Initialize DisasContext */
db->tb = tb;
db->pc_first = tb->pc;
@@ -122,8 +134,21 @@ void translator_loop(const TranslatorOps *ops, DisasContextBase *db,
db->is_jmp = DISAS_TOO_MANY;
break;
}
+
+ /*
+ * Propagate SEGVs from the first instruction to the guest and handle
+ * the rest. This way guest's siginfo_t gets accurate pc and si_addr.
+ */
+ cpu->translator_jmp = true;
}
+ /*
+ * Clear translator_jmp on all ways out of this function, otherwise
+ * instructions that fetch code as part of their operation will be
+ * confused.
+ */
+ cpu->translator_jmp = false;
+
/* Emit code to exit the TB, as indicated by db->is_jmp. */
ops->tb_stop(db, cpu);
gen_tb_end(db->tb, db->num_insns);
diff --git a/include/hw/core/cpu.h b/include/hw/core/cpu.h
index 500503da13..6c1829b7f5 100644
--- a/include/hw/core/cpu.h
+++ b/include/hw/core/cpu.h
@@ -349,6 +349,8 @@ struct CPUState {
int64_t icount_extra;
uint64_t random_seed;
sigjmp_buf jmp_env;
+ bool translator_jmp;
+ sigjmp_buf translator_jmp_env;
QemuMutex work_mutex;
QSIMPLEQ_HEAD(, qemu_work_item) work_list;
diff --git a/linux-user/signal.c b/linux-user/signal.c
index 8d29bfaa6b..f7e77c8d2e 100644
--- a/linux-user/signal.c
+++ b/linux-user/signal.c
@@ -833,6 +833,11 @@ static void host_signal_handler(int host_sig, siginfo_t *info, void *puc)
abi_ptr guest_addr;
bool is_write;
+ /* Translator wants to handle this. */
+ if (helper_retaddr == 1 && cpu->translator_jmp) {
+ siglongjmp(cpu->translator_jmp_env, 1);
+ }
+
host_addr = (uintptr_t)info->si_addr;
/*
--
2.35.3
next prev parent reply other threads:[~2022-08-04 18:49 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-08-04 18:23 [PATCH 0/2] linux-user: Fix siginfo_t contents when jumping to non-readable pages Ilya Leoshkevich
2022-08-04 18:23 ` Ilya Leoshkevich [this message]
2022-08-05 8:50 ` [PATCH 1/2] " Peter Maydell
2022-08-05 10:28 ` Ilya Leoshkevich
2022-08-05 10:55 ` Peter Maydell
2022-08-04 18:23 ` [PATCH 2/2] tests/tcg: Test " Ilya Leoshkevich
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20220804182359.830058-2-iii@linux.ibm.com \
--to=iii@linux.ibm.com \
--cc=borntraeger@linux.ibm.com \
--cc=david@redhat.com \
--cc=eduardo@habkost.net \
--cc=f4bug@amsat.org \
--cc=laurent@vivier.eu \
--cc=marcel.apfelbaum@gmail.com \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=qemu-s390x@nongnu.org \
--cc=richard.henderson@linaro.org \
--cc=wangyanan55@huawei.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).