From: Joelle van Dyne <j@getutm.app>
To: qemu-devel@nongnu.org
Cc: Paolo Bonzini <pbonzini@redhat.com>,
Joelle van Dyne <j@getutm.app>,
Richard Henderson <rth@twiddle.net>
Subject: [PATCH RESEND 2/4] tcg: implement mirror mapped JIT for Linux
Date: Mon, 26 Oct 2020 12:15:38 -0700 [thread overview]
Message-ID: <20201026191540.12434-3-j@getutm.app> (raw)
In-Reply-To: <20201026191540.12434-1-j@getutm.app>
When configured with --enable-debug-tcg on Linux, we enable JIT mirror map
which separates the JIT allocation into a RX-only address and a RW-only
address. The backing is a memfd_create() object.
This allows us to catch bugs where we unintentionally write to the JIT
region. However, the main use case will be for other platforms that require
strict W^X separation to run.
Signed-off-by: Joelle van Dyne <j@getutm.app>
---
configure | 13 ++++++++
meson.build | 1 +
include/exec/exec-all.h | 8 +++++
include/tcg/tcg.h | 18 ++++++++++
accel/tcg/cpu-exec.c | 7 +++-
accel/tcg/translate-all.c | 69 +++++++++++++++++++++++++++++++++++----
6 files changed, 108 insertions(+), 8 deletions(-)
diff --git a/configure b/configure
index 55e07c82dd..71bbe82ac5 100755
--- a/configure
+++ b/configure
@@ -447,6 +447,7 @@ meson=""
ninja=""
skip_meson=no
gettext=""
+mirror_jit="no"
bogus_os="no"
malloc_trim="auto"
@@ -5715,6 +5716,14 @@ if test "$mingw32" = "yes" ; then
done
fi
+# Enable mirror mapping if debug_tcg on supported platform
+
+if test "$debug_tcg" = "yes"; then
+ if test "$linux" = "yes"; then
+ mirror_jit=yes
+ fi
+fi
+
# We can only support ivshmem if we have eventfd
if [ "$eventfd" = "yes" ]; then
ivshmem=yes
@@ -6801,6 +6810,10 @@ if test "$safe_stack" = "yes"; then
echo "CONFIG_SAFESTACK=y" >> $config_host_mak
fi
+if test "$mirror_jit" = "yes"; then
+ echo "CONFIG_MIRROR_JIT=y" >> $config_host_mak
+fi
+
# If we're using a separate build tree, set it up now.
# DIRS are directories which we simply mkdir in the build tree;
# LINKS are things to symlink back into the source tree
diff --git a/meson.build b/meson.build
index 835424999d..0a56fef146 100644
--- a/meson.build
+++ b/meson.build
@@ -2162,6 +2162,7 @@ endif
summary_info += {'thread sanitizer': config_host.has_key('CONFIG_TSAN')}
summary_info += {'rng-none': config_host.has_key('CONFIG_RNG_NONE')}
summary_info += {'Linux keyring': config_host.has_key('CONFIG_SECRET_KEYRING')}
+summary_info += {'mirror JIT': config_host.has_key('CONFIG_MIRROR_JIT')}
summary(summary_info, bool_yn: true)
if not supported_cpus.contains(cpu)
diff --git a/include/exec/exec-all.h b/include/exec/exec-all.h
index 4707ac140c..921767a51b 100644
--- a/include/exec/exec-all.h
+++ b/include/exec/exec-all.h
@@ -519,6 +519,14 @@ struct TranslationBlock {
uintptr_t jmp_list_head;
uintptr_t jmp_list_next[2];
uintptr_t jmp_dest[2];
+
+#if defined(CONFIG_MIRROR_JIT)
+ /*
+ * Store difference to writable mirror
+ * We need this when patching the jump instructions
+ */
+ ptrdiff_t code_rw_mirror_diff;
+#endif
};
extern bool parallel_cpus;
diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h
index 79c5ff8dab..a76c259d1b 100644
--- a/include/tcg/tcg.h
+++ b/include/tcg/tcg.h
@@ -627,6 +627,10 @@ struct TCGContext {
size_t code_gen_buffer_size;
void *code_gen_ptr;
void *data_gen_ptr;
+#if defined(CONFIG_MIRROR_JIT)
+ int code_gen_buffer_fd;
+ ptrdiff_t code_rw_mirror_diff;
+#endif
/* Threshold to flush the translated code buffer. */
void *code_gen_highwater;
@@ -677,6 +681,20 @@ struct TCGContext {
target_ulong gen_insn_data[TCG_MAX_INSNS][TARGET_INSN_START_WORDS];
};
+static inline void *tcg_mirror_ptr_rw(TCGContext *s, const void *rx)
+{
+#if defined(CONFIG_MIRROR_JIT)
+ return (void *)rx + s->code_rw_mirror_diff;
+#else
+ return (void *)rx;
+#endif
+}
+
+static inline tcg_insn_unit *tcg_code_ptr_rw(TCGContext *s, const void *rx)
+{
+ return (tcg_insn_unit *)tcg_mirror_ptr_rw(s, rx);
+}
+
extern TCGContext tcg_init_ctx;
extern __thread TCGContext *tcg_ctx;
extern TCGv_env cpu_env;
diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c
index 58aea605d8..f7eb6fcd77 100644
--- a/accel/tcg/cpu-exec.c
+++ b/accel/tcg/cpu-exec.c
@@ -354,7 +354,12 @@ void tb_set_jmp_target(TranslationBlock *tb, int n, uintptr_t addr)
if (TCG_TARGET_HAS_direct_jump) {
uintptr_t offset = tb->jmp_target_arg[n];
uintptr_t tc_ptr = (uintptr_t)tb->tc.ptr;
- tb_target_set_jmp_target(tc_ptr, tc_ptr + offset, addr);
+#if defined(CONFIG_MIRROR_JIT)
+ uintptr_t wr_addr = tc_ptr + offset + tb->code_rw_mirror_diff;
+#else
+ uintptr_t wr_addr = tc_ptr + offset;
+#endif
+ tb_target_set_jmp_target(tc_ptr, tc_ptr + offset, addr, wr_addr);
} else {
tb->jmp_target_arg[n] = addr;
}
diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
index d76097296d..bfb856ac1a 100644
--- a/accel/tcg/translate-all.c
+++ b/accel/tcg/translate-all.c
@@ -59,6 +59,7 @@
#include "sysemu/cpus.h"
#include "sysemu/cpu-timers.h"
#include "sysemu/tcg.h"
+#include "qemu/memfd.h"
/* #define DEBUG_TB_INVALIDATE */
/* #define DEBUG_TB_FLUSH */
@@ -302,10 +303,12 @@ static target_long decode_sleb128(uint8_t **pp)
static int encode_search(TranslationBlock *tb, uint8_t *block)
{
- uint8_t *highwater = tcg_ctx->code_gen_highwater;
- uint8_t *p = block;
+ const uint8_t *highwater, *start;
+ uint8_t *p;
int i, j, n;
+ highwater = tcg_mirror_ptr_rw(tcg_ctx, tcg_ctx->code_gen_highwater);
+ start = p = tcg_mirror_ptr_rw(tcg_ctx, block);
for (i = 0, n = tb->icount; i < n; ++i) {
target_ulong prev;
@@ -329,7 +332,7 @@ static int encode_search(TranslationBlock *tb, uint8_t *block)
}
}
- return p - block;
+ return p - start;
}
/* The cpu state corresponding to 'searched_pc' is restored.
@@ -1067,12 +1070,29 @@ static inline void *alloc_code_gen_buffer(void)
#else
static inline void *alloc_code_gen_buffer(void)
{
- int prot = PROT_WRITE | PROT_READ | PROT_EXEC;
- int flags = MAP_PRIVATE | MAP_ANONYMOUS;
+ int prot = PROT_READ | PROT_EXEC;
+ int flags = 0;
size_t size = tcg_ctx->code_gen_buffer_size;
+ int fd = -1;
void *buf;
- buf = mmap(NULL, size, prot, flags, -1, 0);
+#if defined(CONFIG_MIRROR_JIT)
+#if defined(CONFIG_LINUX)
+ fd = qemu_memfd_create("tcg-jit", size, false, 0, 0, NULL);
+ if (fd < 0) {
+ return NULL;
+ }
+ tcg_ctx->code_gen_buffer_fd = fd;
+ flags |= MAP_SHARED;
+#else /* defined(CONFIG_LINUX) */
+#error "Mirror JIT unimplemented for this platform."
+#endif /* defined(CONFIG_LINUX) */
+#else /* defined(CONFIG_MIRROR_JIT) */
+ prot |= PROT_WRITE;
+ flags |= MAP_ANONYMOUS | MAP_PRIVATE;
+#endif /* defined(CONFIG_MIRROR_JIT) */
+
+ buf = mmap(NULL, size, prot, flags, fd, 0);
if (buf == MAP_FAILED) {
return NULL;
}
@@ -1118,7 +1138,28 @@ static inline void *alloc_code_gen_buffer(void)
}
#endif /* USE_STATIC_CODE_GEN_BUFFER, WIN32, POSIX */
-static inline void code_gen_alloc(size_t tb_size)
+#if defined(CONFIG_MIRROR_JIT)
+#if defined(CONFIG_LINUX)
+static inline void *alloc_jit_rw_mirror(void)
+{
+ int fd = tcg_ctx->code_gen_buffer_fd;
+ size_t size = tcg_ctx->code_gen_buffer_size;
+ void *buf;
+
+ assert(fd >= 0);
+ buf = mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0);
+ if (buf == MAP_FAILED) {
+ return NULL;
+ }
+
+ return buf;
+}
+#else
+#error "Mirror JIT unimplemented for this platform."
+#endif /* CONFIG_IOS */
+#endif /* CONFIG_MIRROR_JIT */
+
+static inline void code_gen_alloc(size_t tb_size, bool mirror_jit)
{
tcg_ctx->code_gen_buffer_size = size_code_gen_buffer(tb_size);
tcg_ctx->code_gen_buffer = alloc_code_gen_buffer();
@@ -1126,6 +1167,17 @@ static inline void code_gen_alloc(size_t tb_size)
fprintf(stderr, "Could not allocate dynamic translator buffer\n");
exit(1);
}
+#if defined(CONFIG_MIRROR_JIT)
+ void *mirror;
+
+ /* For platforms that need a mirror mapping for code execution */
+ mirror = alloc_jit_rw_mirror();
+ if (mirror == NULL) {
+ fprintf(stderr, "Could not remap code buffer mirror\n");
+ exit(1);
+ }
+ tcg_ctx->code_rw_mirror_diff = mirror - tcg_ctx->code_gen_buffer;
+#endif /* CONFIG_MIRROR_JIT */
}
static bool tb_cmp(const void *ap, const void *bp)
@@ -1721,6 +1773,9 @@ TranslationBlock *tb_gen_code(CPUState *cpu,
cpu_loop_exit(cpu);
}
+#if defined(CONFIG_MIRROR_JIT)
+ tb->code_rw_mirror_diff = tcg_ctx->code_rw_mirror_diff;
+#endif
gen_code_buf = tcg_ctx->code_gen_ptr;
tb->tc.ptr = gen_code_buf;
tb->pc = pc;
--
2.28.0
next prev parent reply other threads:[~2020-10-26 19:18 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-10-26 19:15 [PATCH RESEND 0/4] Mirror map JIT memory for TCG Joelle van Dyne
2020-10-26 19:15 ` [PATCH RESEND 1/4] tcg: add const hints for code pointers Joelle van Dyne
2020-10-26 19:15 ` Joelle van Dyne
2020-10-26 19:15 ` Joelle van Dyne [this message]
2020-10-26 19:15 ` [PATCH RESEND 3/4] tcg: use mirror map JIT in code generation Joelle van Dyne
2020-10-26 19:15 ` Joelle van Dyne
2020-10-26 19:15 ` [PATCH RESEND 4/4] tcg: make mirror mapped JIT optional Joelle van Dyne
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=20201026191540.12434-3-j@getutm.app \
--to=j@getutm.app \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=rth@twiddle.net \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.