Perf uninitialized value fixes
cover Perf uninitialized value fixes
These patches are all warnings that the MSAN (Memory Sanitizer) build of perf has caught.
To build perf with MSAN enabled run:
make -C tools/perf CLANG=1 CC=clang EXTRA_CFLAGS="-fsanitize=memory\
-fsanitize-memory-track-origins"
(The -fsanitizer-memory-track-origins makes the bugs clearer but isn't strictly necessary.)
(Additionally, llvm might have to be installed and clang might have to be specified as the compiler - export CC=/usr/bin/clang).
The patches "Fix util.c use of uninitialized value warning" and "Fix annotate.c use of uninitialized value error" build on top of each other (the changes in Fix util.c use of uninitialized value warning must be made first).
When running the commands provided in the repro instructions, MSAN will generate false positive uninitialized memory errors. This is happening because libc is not MSAN-instrumented. Finding a way to build libc with MSAN will get rid of these false positives and allow the real warnings mentioned in the patches to be shown.
Numfor Mbiziwo-Tiapo (3):
Fix util.c use of uninitialized value warning Fix annotate.c use of uninitialized value error Fix sched-messaging.c use of uninitialized value errors
tools/perf/bench/sched-messaging.c | 3 ++- tools/perf/util/annotate.c | 15 +++++++++++---- tools/perf/util/header.c | 2 +- 3 files changed, 14 insertions(+), 6 deletions(-)
Signature · 1 line
2.22.0.657.g960e92d24f-goog
PATCH 1/3 Fix util.c use of unitialized value warning
When building our local version of perf with MSAN (Memory Sanitizer) and running the perf record command, MSAN throws a use of uninitialized value warning in "tools/perf/util/util.c:333:6".
This warning stems from the "buf" variable being passed into "write". It originated as the variable "ev" with the type union perf_event* defined in the "perf_event__synthesize_attr" function in "tools/perf/util/header.c".
In the "perf_event__synthesize_attr" function they allocate space with a malloc call using ev, then go on to only assign some of the member variables before passing "ev" on as a parameter to the "process" function therefore "ev" contains uninitialized memory. Changing the malloc call to calloc initializes all the members of "ev" which gets rid of the warning.
To reproduce this warning, build perf by running:
make -C tools/perf CLANG=1 CC=clang EXTRA_CFLAGS="-fsanitize=memory\
-fsanitize-memory-track-origins"
(Additionally, llvm might have to be installed and clang might have to be specified as the compiler - export CC=/usr/bin/clang)
then running:tools/perf/perf record -o - ls / | tools/perf/perf --no-pager annotate\
-i - --stdio
Please see the cover letter for why false positive warnings may be generated.
--- tools/perf/util/header.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/util/header.c +1 -1
@@ -3427,7 +3427,7 @@ int perf_event__synthesize_attr(struct perf_tool *tool, size +=sizeof (struct perf_event_header ); size += ids *sizeof (u64); - ev =malloc (size); + ev =calloc (1 , size);if (ev ==NULL )return -ENOMEM ;
Signature · 1 line
2.22.0.657.g960e92d24f-goog
Re: [PATCH 1/3] Fix util.c use of unitialized value warning
Em Wed, Jul 24, 2019 at 04:44:58PM -0700, Numfor Mbiziwo-Tiapo escreveu:
Quoted 9 lines
When building our local version of perf with MSAN (Memory Sanitizer) and running the perf record command, MSAN throws a use of uninitialized value warning in "
tools/perf/util/util.c:333:6".This warning stems from the "buf" variable being passed into "write". It originated as the variable "ev" with the type union perf_event* defined in the "
perf_event__synthesize_attr" function in "tools/perf/util/header.c".
Quoted 6 lines
In the "
perf_event__synthesize_attr" function they allocate space with a malloc call using ev, then go on to only assign some of the member variables before passing "ev" on as a parameter to the "process" function therefore "ev" contains uninitialized memory. Changing the malloc call to calloc initializes all the members of "ev" which gets rid of the warning.
I'm applying, but changing the calloc call to zalloc() that is just a shorter wrapper to calloc(1, size).
Thanks,
- Arnaldo
Quoted 34 lines
To reproduce this warning, build perf by running:
make -Ctools/perfCLANG=1 CC=clang EXTRA_CFLAGS="-fsanitize=memory\
-fsanitize-memory-track-origins"(Additionally, llvm might have to be installed and clang might have to be specified as the compiler - export CC=/usr/bin/clang)
then running:
tools/perf/perfrecord -o - ls / |tools/perf/perf--no-pager annotate\
-i - --stdioPlease see the cover letter for why false positive warnings may be generated.
Signed-off-by Numfor Mbiziwo-Tiapo <[email protected]>--- tools/perf/util/header.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)tools/perf/util/header.c +1
-1@@ -3427,7 +3427,7 @@ int perf_event__synthesize_attr(struct perf_tool *tool, size +=sizeof (struct perf_event_header ); size += ids *sizeof (u64); - ev =malloc (size); + ev =calloc (1 , size);if (ev ==NULL )return -ENOMEM ;Signature · 1 line
2.22.0.657.g960e92d24f-goog
Signature · 1 line
- Arnaldo
tip-bot for Numfor Mbiziwo-Tiapo
merge bot[tip:perf/urgent] perf header: Fix use of unitialized value warning
In reply to Numfor Mbiziwo-Tiapoperf header: Fix use of unitialized value warning
Commit-ID: 20f9781f491360e7459c589705a2e4b1f136bee9 Gitweb: https://git.kernel.org/tip/20f9781f491360e7459c589705a2e4b1f136bee9 Author: Numfor Mbiziwo-Tiapo <[email protected]> AuthorDate: Wed, 24 Jul 2019 16:44:58 -0700 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Mon, 29 Jul 2019 09:03:43 -0300
perf header: Fix use of unitialized value warning
When building our local version of perf with MSAN (Memory Sanitizer) and running the perf record command, MSAN throws a use of uninitialized value warning in "tools/perf/util/util.c:333:6".
This warning stems from the "buf" variable being passed into "write". It originated as the variable "ev" with the type union perf_event* defined in the "perf_event__synthesize_attr" function in "tools/perf/util/header.c".
In the "perf_event__synthesize_attr" function they allocate space with a malloc call using ev, then go on to only assign some of the member variables before passing "ev" on as a parameter to the "process" function therefore "ev" contains uninitialized memory. Changing the malloc call to zalloc to initialize all the members of "ev" which gets rid of the warning.
To reproduce this warning, build perf by running:
make -C tools/perf CLANG=1 CC=clang EXTRA_CFLAGS="-fsanitize=memory\
-fsanitize-memory-track-origins"
(Additionally, llvm might have to be installed and clang might have to be specified as the compiler - export CC=/usr/bin/clang)
then running:tools/perf/perf record -o - ls / | tools/perf/perf --no-pager annotate\
-i - --stdio
Please see the cover letter for why false positive warnings may be generated.
--- tools/perf/util/header.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/util/header.c +1 -1
@@ -3646,7 +3646,7 @@ int perf_event__synthesize_attr(struct perf_tool *tool, size +=sizeof (struct perf_event_header ); size += ids *sizeof (u64); - ev =malloc (size); + ev =zalloc (size);if (ev ==NULL )return -ENOMEM ;
PATCH 2/3 Fix annotate.c use of uninitialized value error
In reply to Numfor Mbiziwo-TiapoOur local MSAN (Memory Sanitizer) build of perf throws a warning that comes from the "dso__disassemble_filename" function in "tools/perf/util/annotate.c" when running perf record.
The warning stems from the call to readlink, in which "build_id_path" was being read into "linkname". Since readlink does not null terminate, an uninitialized memory access would later occur when "linkname" is passed into the strstr function. This is simply fixed by null-terminating "linkname" after the call to readlink.
To reproduce this warning, build perf by running:
make -C tools/perf CLANG=1 CC=clang EXTRA_CFLAGS="-fsanitize=memory\
-fsanitize-memory-track-origins"
(Additionally, llvm might have to be installed and clang might have to be specified as the compiler - export CC=/usr/bin/clang)
then running:tools/perf/perf record -o - ls / | tools/perf/perf --no-pager annotate\
-i - --stdio
Please see the cover letter for why false positive warnings may be generated.
--- tools/perf/util/annotate.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-)
tools/perf/util/annotate.c +11 -4
@@ -1627,6 +1627,7 @@ static int dso__disassemble_filename(struct dso *dso, char *filename, size_t filchar *build_id_filename;char *build_id_path =NULL ;char *pos; +int len;if (dso->symtab_type ==DSO_BINARY_TYPE__KALLSYMS && !dso__is_kcore (dso))
@@ -1655,10 +1656,16 @@ static int dso__disassemble_filename(struct dso *dso, char *filename, size_t filif (pos &&strlen (pos) <SBUILD_ID_SIZE -2 )dirname (build_id_path); -if (dso__is_kcore (dso) || -readlink (build_id_path, linkname,sizeof (linkname)) <0 || -strstr (linkname,DSO__NAME_KALLSYMS ) || -access (filename,R_OK )) { +if (dso__is_kcore (dso)) + goto fallback; + + len =readlink (build_id_path, linkname,sizeof (linkname)); +if (len <0 ) + goto fallback; + + linkname[len] ='\0' ; +if (strstr (linkname,DSO__NAME_KALLSYMS ) || +access (filename,R_OK )) { fallback:/* * If we don't have build-ids or the build-id file isn't in the
Signature · 1 line
2.22.0.657.g960e92d24f-goog
Re: [PATCH 2/3] Fix annotate.c use of uninitialized value error
Em Wed, Jul 24, 2019 at 04:44:59PM -0700, Numfor Mbiziwo-Tiapo escreveu:
Quoted 45 lines
Our local MSAN (Memory Sanitizer) build of perf throws a warning that comes from the "
dso__disassemble_filename" function in "tools/perf/util/annotate.c" when running perf record.The warning stems from the call to readlink, in which "
build_id_path" was being read into "linkname". Since readlink does not null terminate, an uninitialized memory access would later occur when "linkname" is passed into the strstr function. This is simply fixed by null-terminating "linkname" after the call to readlink.To reproduce this warning, build perf by running:
make -Ctools/perfCLANG=1 CC=clang EXTRA_CFLAGS="-fsanitize=memory\
-fsanitize-memory-track-origins"(Additionally, llvm might have to be installed and clang might have to be specified as the compiler - export CC=/usr/bin/clang)
then running:
tools/perf/perfrecord -o - ls / |tools/perf/perf--no-pager annotate\
-i - --stdioPlease see the cover letter for why false positive warnings may be generated.
Signed-off-by Numfor Mbiziwo-Tiapo <[email protected]>--- tools/perf/util/annotate.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-)tools/perf/util/annotate.c +1
-0@@ -1627,6 +1627,7 @@ static int dso__disassemble_filename(struct dso *dso, char *filename, size_t filchar *build_id_filename;char *build_id_path =NULL ;char *pos; +int len;if (dso->symtab_type ==DSO_BINARY_TYPE__KALLSYMS && !dso__is_kcore (dso))@@ -1655,10 +1656,16 @@ static int dso__disassemble_filename(struct dso *dso, char *filename, size_t filif (pos &&strlen (pos) <SBUILD_ID_SIZE -2 )dirname (build_id_path);
tools/perf/util/annotate.c static int dso__disassemble_filename(struct dso *dso, char *filename, size_t fil · line 1664
Quoted 10 lines
tools/perf/util/annotate.c +6
-4-if (dso__is_kcore (dso) || -readlink (build_id_path, linkname,sizeof (linkname)) <0 || -strstr (linkname,DSO__NAME_KALLSYMS ) || -access (filename,R_OK )) { +if (dso__is_kcore (dso)) + goto fallback; + + len =readlink (build_id_path, linkname,sizeof (linkname)); +if (len <0 ) + goto fallback;
If the buffer has N bytes and the linkname has more than that, then it will truncate at N bytes and return N, then if we access linkname[N] we will be accessing one byte after the end of the buffer, no?
Quoted 9 lines
tools/perf/util/annotate.c +4
-0+ + linkname[len] ='\0' ; +if (strstr (linkname,DSO__NAME_KALLSYMS ) || +access (filename,R_OK )) { fallback:/* * If we don't have build-ids or the build-id file isn't in the Signature · 1 line
2.22.0.657.g960e92d24f-goog
Signature · 1 line
- Arnaldo
PATCH v2 1/1 Fix annotate.c use of uninitialized value error
Our local MSAN (Memory Sanitizer) build of perf throws a warning that comes from the "dso__disassemble_filename" function in "tools/perf/util/annotate.c" when running perf record.
The warning stems from the call to readlink, in which "build_id_path" was being read into "linkname". Since readlink does not null terminate, an uninitialized memory access would later occur when "linkname" is passed into the strstr function. This is simply fixed by null-terminating "linkname" after the call to readlink.
To reproduce this warning, build perf by running:
make -C tools/perf CLANG=1 CC=clang EXTRA_CFLAGS="-fsanitize=memory\
-fsanitize-memory-track-origins"
(Additionally, llvm might have to be installed and clang might have to be specified as the compiler - export CC=/usr/bin/clang)
then running:tools/perf/perf record -o - ls / | tools/perf/perf --no-pager annotate\
-i - --stdio
Please see the cover letter for why false positive warnings may be generated.
--- tools/perf/util/annotate.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-)
tools/perf/util/annotate.c +11 -4
@@ -1627,6 +1627,7 @@ static int dso__disassemble_filename(struct dso *dso, char *filename, size_t filchar *build_id_filename;char *build_id_path =NULL ;char *pos; +int len;if (dso->symtab_type ==DSO_BINARY_TYPE__KALLSYMS && !dso__is_kcore (dso))
@@ -1655,10 +1656,16 @@ static int dso__disassemble_filename(struct dso *dso, char *filename, size_t filif (pos &&strlen (pos) <SBUILD_ID_SIZE -2 )dirname (build_id_path); -if (dso__is_kcore (dso) || -readlink (build_id_path, linkname,sizeof (linkname)) <0 || -strstr (linkname,DSO__NAME_KALLSYMS ) || -access (filename,R_OK )) { +if (dso__is_kcore (dso)) + goto fallback; + + len =readlink (build_id_path, linkname,sizeof (linkname) -1 ); +if (len <0 ) + goto fallback; + + linkname[len] ='\0' ; +if (strstr (linkname,DSO__NAME_KALLSYMS ) || +access (filename,R_OK )) { fallback:/* * If we don't have build-ids or the build-id file isn't in the
Signature · 1 line
2.22.0.709.g102302147b-goog
PATCH 3/3 Fix sched-messaging.c use of uninitialized value errors
In reply to Numfor Mbiziwo-TiapoOur local MSAN (Memory Sanitizer) build of perf throws use of uninitialized value warnings in "tools/perf/bench/sched-messaging.c" when running perf bench.
The first warning comes from the "ready" function where the "dummy" char is declared and then passed into "write" without being initialized. Initializing "dummy" to any character silences the warning.
The second warning comes from the "sender" function where a "write" call is made to write the contents from the "data" char array when it has not yet been initialized. Calling memset on "data" silences the warning.
To reproduce this warning, build perf by running:
make -C tools/perf CLANG=1 CC=clang EXTRA_CFLAGS="-fsanitize=memory\
-fsanitize-memory-track-origins"
(Additionally, llvm might have to be installed and clang might have to be specified as the compiler - export CC=/usr/bin/clang)
then running: tools/perf/perf bench sched all
Please see the cover letter for why false positive warnings may be generated.
--- tools/perf/bench/sched-messaging.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
tools/perf/bench/sched-messaging.c +2 -1
@@ -69,7 +69,7 @@ static void fdpair(int fds[2])/* Block until we're ready to go */ static void ready (int ready_out,int wakefd) { -char dummy; +char dummy ='N' ;struct pollfd pollfd = { .fd = wakefd, .events =POLLIN };/* Tell them we're ready. */
@@ -87,6 +87,7 @@ static void *sender(struct sender_context *ctx)char data[DATASIZE ];unsigned int i, j; +memset (data,'N' ,DATASIZE );ready (ctx->ready_out, ctx->wakefd);/* Now pump to every receiver. */
Signature · 1 line
2.22.0.657.g960e92d24f-goog
Re: [PATCH 3/3] Fix sched-messaging.c use of uninitialized value errors
Em Wed, Jul 24, 2019 at 04:45:00PM -0700, Numfor Mbiziwo-Tiapo escreveu:
Quoted 11 lines
Our local MSAN (Memory Sanitizer) build of perf throws use of uninitialized value warnings in "
tools/perf/bench/sched-messaging.c" when running perf bench.The first warning comes from the "ready" function where the "dummy" char is declared and then passed into "write" without being initialized. Initializing "dummy" to any character silences the warning.
The second warning comes from the "sender" function where a "write" call is made to write the contents from the "data" char array when it has not yet been initialized. Calling memset on "data" silences the warning.
So, this is just to silence MSAN, as it doesn't matter what is sent, whatever values are in those variables is ok, as it will not be used, right?
- Arnaldo
Quoted 40 lines
To reproduce this warning, build perf by running:
make -Ctools/perfCLANG=1 CC=clang EXTRA_CFLAGS="-fsanitize=memory\
-fsanitize-memory-track-origins"(Additionally, llvm might have to be installed and clang might have to be specified as the compiler - export CC=/usr/bin/clang)
then running:
tools/perf/perfbench sched allPlease see the cover letter for why false positive warnings may be generated.
Signed-off-by Numfor Mbiziwo-Tiapo <[email protected]>--- tools/perf/bench/sched-messaging.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)tools/perf/bench/sched-messaging.c +2
-1@@ -69,7 +69,7 @@ static void fdpair(int fds[2])/* Block until we're ready to go */ static void ready (int ready_out,int wakefd) { -char dummy; +char dummy ='N' ;struct pollfd pollfd = { .fd = wakefd, .events =POLLIN };/* Tell them we're ready. */ @@ -87,6 +87,7 @@ static void *sender(struct sender_context *ctx)char data[DATASIZE ];unsigned int i, j; +memset (data,'N' ,DATASIZE );ready (ctx->ready_out, ctx->wakefd);/* Now pump to every receiver. */ Signature · 1 line
2.22.0.657.g960e92d24f-goog
Signature · 1 line
- Arnaldo
Ian Rogers
reviewerRe: [PATCH 3/3] Fix sched-messaging.c use of uninitialized value errors
Quoted 10 lines · Arnaldo Carvalho de Melo
Em Wed, Jul 24, 2019 at 04:45:00PM -0700, Numfor Mbiziwo-Tiapo escreveu:
Quoted 8 lines
Our local MSAN (Memory Sanitizer) build of perf throws use of uninitialized value warnings in "
tools/perf/bench/sched-messaging.c" when running perf bench.The first warning comes from the "ready" function where the "dummy" char is declared and then passed into "write" without being initialized. Initializing "dummy" to any character silences the warning.
Quoted 7 lines
Quoted 3 lines
The second warning comes from the "sender" function where a "write" call is made to write the contents from the "data" char array when it has not yet been initialized. Calling memset on "data" silences the warning.
So, this is just to silence MSAN, as it doesn't matter what is sent, whatever values are in those variables is ok, as it will not be used, right?
That's right.
Thanks,
Ian Rogers
Quoted 46 lines
- Arnaldo
Quoted 39 lines
To reproduce this warning, build perf by running:
make -Ctools/perfCLANG=1 CC=clang EXTRA_CFLAGS="-fsanitize=memory\
-fsanitize-memory-track-origins"(Additionally, llvm might have to be installed and clang might have to be specified as the compiler - export CC=/usr/bin/clang)
then running:
tools/perf/perfbench sched allPlease see the cover letter for why false positive warnings may be generated.
Signed-off-by Numfor Mbiziwo-Tiapo <[email protected]>--- tools/perf/bench/sched-messaging.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)tools/perf/bench/sched-messaging.c +2
-2@@ -69,7 +69,7 @@ static void fdpair(int fds[2])/* Block until we're ready to go */ static void ready (int ready_out,int wakefd) { -char dummy; +char dummy ='N' ;struct pollfd pollfd = { .fd = wakefd, .events =POLLIN };/* Tell them we're ready. */ @@ -87,6 +87,7 @@ static void *sender(struct sender_context *ctx)char data[DATASIZE ];unsigned int i, j; +memset (data,'N' ,DATASIZE );ready (ctx->ready_out, ctx->wakefd);/* Now pump to every receiver. */ --Quoted 1 line
2.22.0.657.g960e92d24f-goog
--
- Arnaldo
Em Wed, Jul 24, 2019 at 04:44:57PM -0700, Numfor Mbiziwo-Tiapo escreveu:
Quoted 18 lines
These patches are all warnings that the MSAN (Memory Sanitizer) build of perf has caught.
To build perf with MSAN enabled run:
make -Ctools/perfCLANG=1 CC=clang EXTRA_CFLAGS="-fsanitize=memory\
-fsanitize-memory-track-origins"(The -fsanitizer-memory-track-origins makes the bugs clearer but isn't strictly necessary.)
(Additionally, llvm might have to be installed and clang might have to be specified as the compiler - export CC=/usr/bin/clang).
The patches "Fix
util.cuse of uninitialized value warning" and "Fixannotate.cuse of uninitialized value error" build on top of each other (the changes in Fixutil.cuse of uninitialized value warning must be made first).
Quoted 5 lines
When running the commands provided in the repro instructions, MSAN will generate false positive uninitialized memory errors. This is happening because libc is not MSAN-instrumented. Finding a way to build libc with MSAN will get rid of these false positives and allow the real warnings mentioned in the patches to be shown.
So this is because I'm not running a glibc linked with MSAN? Do you have any pointer to help building glibc with MSAN? I want to do that inside a container so that I can use these sanitizers, thanks,
[root@quaco ~]# perf record -o - ls / | perf --no-pager annotate -i - --stdio ==29732==WARNING: MemorySanitizer: use-of-uninitialized-value ==29733==WARNING: MemorySanitizer: use-of-uninitialized-value
#0 0xcc136d in add_path /home/acme/git/perf/tools/lib/subcmd/exec-cmd.c:130:6
#1 0xcc075e in setup_path /home/acme/git/perf/tools/lib/subcmd/exec-cmd.c:146:2
#2 0x71298d in main /home/acme/git/perf/tools/perf/perf.c:512:2
#0 0xcc136d in add_path /home/acme/git/perf/tools/lib/subcmd/exec-cmd.c:130:6
#1 0xcc075e in setup_path /home/acme/git/perf/tools/lib/subcmd/exec-cmd.c:146:2
#2 0x71298d in main /home/acme/git/perf/tools/perf/perf.c:512:2
#3 0x7f45b9e29f32 in __libc_start_main (/lib64/libc.so.6+0x23f32)
#4 0x447dcd in _start (/home/acme/bin/perf+0x447dcd)Uninitialized value was created by a heap allocation
#3 0x7fd6433cff32 in __libc_start_main (/lib64/libc.so.6+0x23f32)
#4 0x447dcd in _start (/home/acme/bin/perf+0x447dcd)Uninitialized value was created by a heap allocation
#0 0x4507d2 in malloc /home/acme/git/llvm/projects/compiler-rt/lib/msan/msan_interceptors.cc:916:3
#1 0x7f45b9e7fc47 in __vasprintf_internal (/lib64/libc.so.6+0x79c47)SUMMARY: MemorySanitizer: use-of-uninitialized-value /home/acme/git/perf/tools/lib/subcmd/exec-cmd.c:130:6 in add_path Exiting
#0 0x4507d2 in malloc /home/acme/git/llvm/projects/compiler-rt/lib/msan/msan_interceptors.cc:916:3
#1 0x7fd643425c47 in __vasprintf_internal (/lib64/libc.so.6+0x79c47)SUMMARY: MemorySanitizer: use-of-uninitialized-value /home/acme/git/perf/tools/lib/subcmd/exec-cmd.c:130:6 in add_path Exiting
[root@quaco ~]#
Quoted 12 lines
Numfor Mbiziwo-Tiapo (3):
Fixutil.cuse of uninitialized value warning Fixannotate.cuse of uninitialized value error Fixsched-messaging.cuse of uninitialized value errorstools/perf/bench/sched-messaging.c | 3 ++- tools/perf/util/annotate.c | 15 +++++++++++---- tools/perf/util/header.c | 2 +- 3 files changed, 14 insertions(+), 6 deletions(-)Signature · 1 line
2.22.0.657.g960e92d24f-goog
Signature · 1 line
- Arnaldo
Ian Rogers
reviewerQuoted 50 lines · Arnaldo Carvalho de Melo
Em Wed, Jul 24, 2019 at 04:44:57PM -0700, Numfor Mbiziwo-Tiapo escreveu:
Quoted 18 lines
These patches are all warnings that the MSAN (Memory Sanitizer) build of perf has caught.
To build perf with MSAN enabled run:
make -Ctools/perfCLANG=1 CC=clang EXTRA_CFLAGS="-fsanitize=memory\
-fsanitize-memory-track-origins"(The -fsanitizer-memory-track-origins makes the bugs clearer but isn't strictly necessary.)
(Additionally, llvm might have to be installed and clang might have to be specified as the compiler - export CC=/usr/bin/clang).
The patches "Fix
util.cuse of uninitialized value warning" and "Fixannotate.cuse of uninitialized value error" build on top of each other (the changes in Fixutil.cuse of uninitialized value warning must be made first).Quoted 5 lines
When running the commands provided in the repro instructions, MSAN will generate false positive uninitialized memory errors. This is happening because libc is not MSAN-instrumented. Finding a way to build libc with MSAN will get rid of these false positives and allow the real warnings mentioned in the patches to be shown.
So this is because I'm not running a glibc linked with MSAN? Do you have any pointer to help building glibc with MSAN? I want to do that inside a container so that I can use these sanitizers, thanks,
[root@quaco ~]# perf record -o - ls / | perf --no-pager annotate -i - --stdio ==29732==WARNING: MemorySanitizer: use-of-uninitialized-value ==29733==WARNING: MemorySanitizer: use-of-uninitialized-value
#0 0xcc136d in add_path /home/acme/git/perf/tools/lib/subcmd/exec-cmd.c:130:6 #1 0xcc075e in setup_path /home/acme/git/perf/tools/lib/subcmd/exec-cmd.c:146:2 #2 0x71298d in main /home/acme/git/perf/tools/perf/perf.c:512:2 #0 0xcc136d in add_path /home/acme/git/perf/tools/lib/subcmd/exec-cmd.c:130:6 #1 0xcc075e in setup_path /home/acme/git/perf/tools/lib/subcmd/exec-cmd.c:146:2 #2 0x71298d in main /home/acme/git/perf/tools/perf/perf.c:512:2 #3 0x7f45b9e29f32 in __libc_start_main (/lib64/libc.so.6+0x23f32) #4 0x447dcd in _start (/home/acme/bin/perf+0x447dcd)Uninitialized value was created by a heap allocation
#3 0x7fd6433cff32 in __libc_start_main (/lib64/libc.so.6+0x23f32) #4 0x447dcd in _start (/home/acme/bin/perf+0x447dcd)Uninitialized value was created by a heap allocation
#0 0x4507d2 in malloc /home/acme/git/llvm/projects/compiler-rt/lib/msan/msan_interceptors.cc:916:3 #1 0x7f45b9e7fc47 in __vasprintf_internal (/lib64/libc.so.6+0x79c47)
Quoted 18 lines
SUMMARY: MemorySanitizer: use-of-uninitialized-value
/home/acme/git/perf/tools/lib/subcmd/exec-cmd.c:130:6inadd_pathExiting#0 0x4507d2 in malloc /home/acme/git/llvm/projects/compiler-rt/lib/msan/msan_interceptors.cc:916:3 #1 0x7fd643425c47 in __vasprintf_internal (/lib64/libc.so.6+0x79c47)SUMMARY: MemorySanitizer: use-of-uninitialized-value
/home/acme/git/perf/tools/lib/subcmd/exec-cmd.c:130:6inadd_pathExiting
[root@quaco ~]#Quoted 9 lines
Numfor Mbiziwo-Tiapo (3):
Fixutil.cuse of uninitialized value warning Fixannotate.cuse of uninitialized value error Fixsched-messaging.cuse of uninitialized value errorstools/perf/bench/sched-messaging.c | 3 ++- tools/perf/util/annotate.c | 15 +++++++++++---- tools/perf/util/header.c | 2 +- 3 files changed, 14 insertions(+), 6 deletions(-)
Thanks Arnaldo! Debugging the issue it isn't down glibc, there are interceptors in the sanitizers for asprintf to recognize it as a source of memory allocation. The problem is the sanitizers don't support _FORTIFY_SOURCE [1] and this is causing the false positives. The following patch works to resolve the false-positive issue for me:
-----
tools/lib/subcmd/Makefile +0 -0
@@ -20,7 +20,13 @@ MAKEFLAGS += --no-print-directory
LIBFILE = $(OUTPUT)libsubcmd.a
CFLAGS := $(EXTRA_WARNINGS) $(EXTRA_CFLAGS) -CFLAGS += -ggdb3 -Wall -Wextra -std=gnu99 -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2 -fPIC
+CFLAGS += -ggdb3 -Wall -Wextra -std=gnu99 -fPIC
+
+ifeq ($(DEBUG),0)
+ ifeq ($(feature-fortify-source), 1)
+ CFLAGS += -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2
+ endif
+endif
ifeq ($(CC_NO_CLANG), 0)
CFLAGS += -O3
-----
Thanks,
Ian
[1] https://github.com/google/sanitizers/wiki/AddressSanitizer#faq
Quoted 6 lines
Quoted 2 lines
--
2.22.0.657.g960e92d24f-goog--
- Arnaldo