On 7/17/21 19:55, Mathy Vanhoef wrote:
> On 7/16/21 11:11 AM, Zheng Yejian wrote:
> > In v4.4, commit e76511a6fbb5 ("mac80211: properly handle A-MSDUs that
> > start with an RFC 1042 header") looks like an incomplete backport.
> >
> > There is no functional changes in the commit, since
> > __ieee80211_data_to_8023() which defined in net/wireless/util.c is
> > only called by ieee80211_data_to_8023() and parameter 'is_amsdu' is
> > always input as false.
>
> I don't think there's a problem here. The core commit that prevents the
> A-MSDU attack is "[PATCH 04/18] cfg80211: mitigate A-MSDU aggregation
> attacks":
> https://lore.kernel.org/linux-
> wireless/20210511200110.25d93176ddaf.I9e265b597f2cd23eb44573f35b62594
> 7b386a9de@changeid/
>
> That commit states: "for kernel 4.9 and above this patch depends on
> "mac80211: properly handle A-MSDUs that start with a rfc1042 header".
> Otherwise this patch has no impact and attacks will remain possible."
>
> Put differently, when patching v4.4 there was in fact no need to
> backport the patch that we're discussing here. So it makes sense that
> the "backported" patches causes no functional changes.
>
> Section 3.6 of https://papers.mathyvanhoef.com/usenix2021.pdf briefly
> discusses the wrong behavior of Linux 4.9+ that this patch tries to fix:
> "Linux 4.9 and above .. strip away the first 8 bytes of an A-MSDU frame
> if these bytes look like a valid LLC/SNAP header, and then further
> process the frame. This behavior is not compliant with the 802.11 standard."
>
How about linux 4.9 below, are they compliant with 802.11 standard or not?
Would they need additional patches to mitigate the aggregation attack?
I know little about 802.11 standard, sorry for that : (
> That said, I didn't yet run the test tool against a patched 4.4 kernel,
> so I hope my understanding of this code in this version is correct.
>
> Best regards,
> Mathy
Thanks,
Zheng Yejian
The patch below does not apply to the 5.4-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable(a)vger.kernel.org>.
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
>From 892ba7f18621a02af4428c58d97451f64685dba4 Mon Sep 17 00:00:00 2001
From: Namhyung Kim <namhyung(a)kernel.org>
Date: Tue, 29 Jun 2021 21:30:58 -0700
Subject: [PATCH] perf report: Fix --task and --stat with pipe input
Current 'perf report' fails to process a pipe input when --task or
--stat options are used. This is because they reset all the tool
callbacks and fails to find a matching event for a sample.
When pipe input is used, the event info is passed via ATTR records so it
needs to handle that operation. Otherwise the following error occurs.
Note, -14 (= -EFAULT) comes from evlist__parse_sample():
# perf record -a -o- sleep 1 | perf report -i- --stat
Can't parse sample, err = -14
0x271044 [0x38]: failed to process type: 9
Error:
failed to process sample
#
Committer testing:
Before:
$ perf record -o- sleep 1 | perf report -i- --stat
Can't parse sample, err = -14
[ perf record: Woken up 1 times to write data ]
0x1350 [0x30]: failed to process type: 9
Error:
failed to process sample
[ perf record: Captured and wrote 0.000 MB - ]
$
After:
$ perf record -o- sleep 1 | perf report -i- --stat
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.000 MB - ]
Aggregated stats:
TOTAL events: 41
COMM events: 2 ( 4.9%)
EXIT events: 1 ( 2.4%)
SAMPLE events: 9 (22.0%)
MMAP2 events: 4 ( 9.8%)
ATTR events: 1 ( 2.4%)
FINISHED_ROUND events: 1 ( 2.4%)
THREAD_MAP events: 1 ( 2.4%)
CPU_MAP events: 1 ( 2.4%)
EVENT_UPDATE events: 1 ( 2.4%)
TIME_CONV events: 1 ( 2.4%)
FEATURE events: 19 (46.3%)
cycles:uhH stats:
SAMPLE events: 9
$
Fixes: a4a4d0a7a2b20f78 ("perf report: Add --stats option to display quick data statistics")
Signed-off-by: Namhyung Kim <namhyung(a)kernel.org>
Acked-by: Jiri Olsa <jolsa(a)redhat.com>
Tested-by: Arnaldo Carvalho de Melo <acme(a)redhat.com>
Cc: Andi Kleen <ak(a)linux.intel.com>
Cc: Ian Rogers <irogers(a)google.com>
Cc: Peter Zijlstra <peterz(a)infradead.org>
Link: http://lore.kernel.org/lkml/20210630043058.1131295-1-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme(a)redhat.com>
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index bc5c393021dc..8639bbe0969d 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -729,9 +729,14 @@ static int count_sample_event(struct perf_tool *tool __maybe_unused,
return 0;
}
+static int process_attr(struct perf_tool *tool __maybe_unused,
+ union perf_event *event,
+ struct evlist **pevlist);
+
static void stats_setup(struct report *rep)
{
memset(&rep->tool, 0, sizeof(rep->tool));
+ rep->tool.attr = process_attr;
rep->tool.sample = count_sample_event;
rep->tool.no_warn = true;
}
@@ -753,6 +758,7 @@ static void tasks_setup(struct report *rep)
rep->tool.mmap = perf_event__process_mmap;
rep->tool.mmap2 = perf_event__process_mmap2;
}
+ rep->tool.attr = process_attr;
rep->tool.comm = perf_event__process_comm;
rep->tool.exit = perf_event__process_exit;
rep->tool.fork = perf_event__process_fork;
The patch below does not apply to the 4.19-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable(a)vger.kernel.org>.
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
>From 892ba7f18621a02af4428c58d97451f64685dba4 Mon Sep 17 00:00:00 2001
From: Namhyung Kim <namhyung(a)kernel.org>
Date: Tue, 29 Jun 2021 21:30:58 -0700
Subject: [PATCH] perf report: Fix --task and --stat with pipe input
Current 'perf report' fails to process a pipe input when --task or
--stat options are used. This is because they reset all the tool
callbacks and fails to find a matching event for a sample.
When pipe input is used, the event info is passed via ATTR records so it
needs to handle that operation. Otherwise the following error occurs.
Note, -14 (= -EFAULT) comes from evlist__parse_sample():
# perf record -a -o- sleep 1 | perf report -i- --stat
Can't parse sample, err = -14
0x271044 [0x38]: failed to process type: 9
Error:
failed to process sample
#
Committer testing:
Before:
$ perf record -o- sleep 1 | perf report -i- --stat
Can't parse sample, err = -14
[ perf record: Woken up 1 times to write data ]
0x1350 [0x30]: failed to process type: 9
Error:
failed to process sample
[ perf record: Captured and wrote 0.000 MB - ]
$
After:
$ perf record -o- sleep 1 | perf report -i- --stat
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.000 MB - ]
Aggregated stats:
TOTAL events: 41
COMM events: 2 ( 4.9%)
EXIT events: 1 ( 2.4%)
SAMPLE events: 9 (22.0%)
MMAP2 events: 4 ( 9.8%)
ATTR events: 1 ( 2.4%)
FINISHED_ROUND events: 1 ( 2.4%)
THREAD_MAP events: 1 ( 2.4%)
CPU_MAP events: 1 ( 2.4%)
EVENT_UPDATE events: 1 ( 2.4%)
TIME_CONV events: 1 ( 2.4%)
FEATURE events: 19 (46.3%)
cycles:uhH stats:
SAMPLE events: 9
$
Fixes: a4a4d0a7a2b20f78 ("perf report: Add --stats option to display quick data statistics")
Signed-off-by: Namhyung Kim <namhyung(a)kernel.org>
Acked-by: Jiri Olsa <jolsa(a)redhat.com>
Tested-by: Arnaldo Carvalho de Melo <acme(a)redhat.com>
Cc: Andi Kleen <ak(a)linux.intel.com>
Cc: Ian Rogers <irogers(a)google.com>
Cc: Peter Zijlstra <peterz(a)infradead.org>
Link: http://lore.kernel.org/lkml/20210630043058.1131295-1-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme(a)redhat.com>
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index bc5c393021dc..8639bbe0969d 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -729,9 +729,14 @@ static int count_sample_event(struct perf_tool *tool __maybe_unused,
return 0;
}
+static int process_attr(struct perf_tool *tool __maybe_unused,
+ union perf_event *event,
+ struct evlist **pevlist);
+
static void stats_setup(struct report *rep)
{
memset(&rep->tool, 0, sizeof(rep->tool));
+ rep->tool.attr = process_attr;
rep->tool.sample = count_sample_event;
rep->tool.no_warn = true;
}
@@ -753,6 +758,7 @@ static void tasks_setup(struct report *rep)
rep->tool.mmap = perf_event__process_mmap;
rep->tool.mmap2 = perf_event__process_mmap2;
}
+ rep->tool.attr = process_attr;
rep->tool.comm = perf_event__process_comm;
rep->tool.exit = perf_event__process_exit;
rep->tool.fork = perf_event__process_fork;
The patch below does not apply to the 5.13-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable(a)vger.kernel.org>.
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
>From 892ba7f18621a02af4428c58d97451f64685dba4 Mon Sep 17 00:00:00 2001
From: Namhyung Kim <namhyung(a)kernel.org>
Date: Tue, 29 Jun 2021 21:30:58 -0700
Subject: [PATCH] perf report: Fix --task and --stat with pipe input
Current 'perf report' fails to process a pipe input when --task or
--stat options are used. This is because they reset all the tool
callbacks and fails to find a matching event for a sample.
When pipe input is used, the event info is passed via ATTR records so it
needs to handle that operation. Otherwise the following error occurs.
Note, -14 (= -EFAULT) comes from evlist__parse_sample():
# perf record -a -o- sleep 1 | perf report -i- --stat
Can't parse sample, err = -14
0x271044 [0x38]: failed to process type: 9
Error:
failed to process sample
#
Committer testing:
Before:
$ perf record -o- sleep 1 | perf report -i- --stat
Can't parse sample, err = -14
[ perf record: Woken up 1 times to write data ]
0x1350 [0x30]: failed to process type: 9
Error:
failed to process sample
[ perf record: Captured and wrote 0.000 MB - ]
$
After:
$ perf record -o- sleep 1 | perf report -i- --stat
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.000 MB - ]
Aggregated stats:
TOTAL events: 41
COMM events: 2 ( 4.9%)
EXIT events: 1 ( 2.4%)
SAMPLE events: 9 (22.0%)
MMAP2 events: 4 ( 9.8%)
ATTR events: 1 ( 2.4%)
FINISHED_ROUND events: 1 ( 2.4%)
THREAD_MAP events: 1 ( 2.4%)
CPU_MAP events: 1 ( 2.4%)
EVENT_UPDATE events: 1 ( 2.4%)
TIME_CONV events: 1 ( 2.4%)
FEATURE events: 19 (46.3%)
cycles:uhH stats:
SAMPLE events: 9
$
Fixes: a4a4d0a7a2b20f78 ("perf report: Add --stats option to display quick data statistics")
Signed-off-by: Namhyung Kim <namhyung(a)kernel.org>
Acked-by: Jiri Olsa <jolsa(a)redhat.com>
Tested-by: Arnaldo Carvalho de Melo <acme(a)redhat.com>
Cc: Andi Kleen <ak(a)linux.intel.com>
Cc: Ian Rogers <irogers(a)google.com>
Cc: Peter Zijlstra <peterz(a)infradead.org>
Link: http://lore.kernel.org/lkml/20210630043058.1131295-1-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme(a)redhat.com>
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index bc5c393021dc..8639bbe0969d 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -729,9 +729,14 @@ static int count_sample_event(struct perf_tool *tool __maybe_unused,
return 0;
}
+static int process_attr(struct perf_tool *tool __maybe_unused,
+ union perf_event *event,
+ struct evlist **pevlist);
+
static void stats_setup(struct report *rep)
{
memset(&rep->tool, 0, sizeof(rep->tool));
+ rep->tool.attr = process_attr;
rep->tool.sample = count_sample_event;
rep->tool.no_warn = true;
}
@@ -753,6 +758,7 @@ static void tasks_setup(struct report *rep)
rep->tool.mmap = perf_event__process_mmap;
rep->tool.mmap2 = perf_event__process_mmap2;
}
+ rep->tool.attr = process_attr;
rep->tool.comm = perf_event__process_comm;
rep->tool.exit = perf_event__process_exit;
rep->tool.fork = perf_event__process_fork;
The patch below does not apply to the 5.12-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable(a)vger.kernel.org>.
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
>From 892ba7f18621a02af4428c58d97451f64685dba4 Mon Sep 17 00:00:00 2001
From: Namhyung Kim <namhyung(a)kernel.org>
Date: Tue, 29 Jun 2021 21:30:58 -0700
Subject: [PATCH] perf report: Fix --task and --stat with pipe input
Current 'perf report' fails to process a pipe input when --task or
--stat options are used. This is because they reset all the tool
callbacks and fails to find a matching event for a sample.
When pipe input is used, the event info is passed via ATTR records so it
needs to handle that operation. Otherwise the following error occurs.
Note, -14 (= -EFAULT) comes from evlist__parse_sample():
# perf record -a -o- sleep 1 | perf report -i- --stat
Can't parse sample, err = -14
0x271044 [0x38]: failed to process type: 9
Error:
failed to process sample
#
Committer testing:
Before:
$ perf record -o- sleep 1 | perf report -i- --stat
Can't parse sample, err = -14
[ perf record: Woken up 1 times to write data ]
0x1350 [0x30]: failed to process type: 9
Error:
failed to process sample
[ perf record: Captured and wrote 0.000 MB - ]
$
After:
$ perf record -o- sleep 1 | perf report -i- --stat
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.000 MB - ]
Aggregated stats:
TOTAL events: 41
COMM events: 2 ( 4.9%)
EXIT events: 1 ( 2.4%)
SAMPLE events: 9 (22.0%)
MMAP2 events: 4 ( 9.8%)
ATTR events: 1 ( 2.4%)
FINISHED_ROUND events: 1 ( 2.4%)
THREAD_MAP events: 1 ( 2.4%)
CPU_MAP events: 1 ( 2.4%)
EVENT_UPDATE events: 1 ( 2.4%)
TIME_CONV events: 1 ( 2.4%)
FEATURE events: 19 (46.3%)
cycles:uhH stats:
SAMPLE events: 9
$
Fixes: a4a4d0a7a2b20f78 ("perf report: Add --stats option to display quick data statistics")
Signed-off-by: Namhyung Kim <namhyung(a)kernel.org>
Acked-by: Jiri Olsa <jolsa(a)redhat.com>
Tested-by: Arnaldo Carvalho de Melo <acme(a)redhat.com>
Cc: Andi Kleen <ak(a)linux.intel.com>
Cc: Ian Rogers <irogers(a)google.com>
Cc: Peter Zijlstra <peterz(a)infradead.org>
Link: http://lore.kernel.org/lkml/20210630043058.1131295-1-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme(a)redhat.com>
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index bc5c393021dc..8639bbe0969d 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -729,9 +729,14 @@ static int count_sample_event(struct perf_tool *tool __maybe_unused,
return 0;
}
+static int process_attr(struct perf_tool *tool __maybe_unused,
+ union perf_event *event,
+ struct evlist **pevlist);
+
static void stats_setup(struct report *rep)
{
memset(&rep->tool, 0, sizeof(rep->tool));
+ rep->tool.attr = process_attr;
rep->tool.sample = count_sample_event;
rep->tool.no_warn = true;
}
@@ -753,6 +758,7 @@ static void tasks_setup(struct report *rep)
rep->tool.mmap = perf_event__process_mmap;
rep->tool.mmap2 = perf_event__process_mmap2;
}
+ rep->tool.attr = process_attr;
rep->tool.comm = perf_event__process_comm;
rep->tool.exit = perf_event__process_exit;
rep->tool.fork = perf_event__process_fork;
The patch below does not apply to the 5.10-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable(a)vger.kernel.org>.
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
>From 892ba7f18621a02af4428c58d97451f64685dba4 Mon Sep 17 00:00:00 2001
From: Namhyung Kim <namhyung(a)kernel.org>
Date: Tue, 29 Jun 2021 21:30:58 -0700
Subject: [PATCH] perf report: Fix --task and --stat with pipe input
Current 'perf report' fails to process a pipe input when --task or
--stat options are used. This is because they reset all the tool
callbacks and fails to find a matching event for a sample.
When pipe input is used, the event info is passed via ATTR records so it
needs to handle that operation. Otherwise the following error occurs.
Note, -14 (= -EFAULT) comes from evlist__parse_sample():
# perf record -a -o- sleep 1 | perf report -i- --stat
Can't parse sample, err = -14
0x271044 [0x38]: failed to process type: 9
Error:
failed to process sample
#
Committer testing:
Before:
$ perf record -o- sleep 1 | perf report -i- --stat
Can't parse sample, err = -14
[ perf record: Woken up 1 times to write data ]
0x1350 [0x30]: failed to process type: 9
Error:
failed to process sample
[ perf record: Captured and wrote 0.000 MB - ]
$
After:
$ perf record -o- sleep 1 | perf report -i- --stat
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.000 MB - ]
Aggregated stats:
TOTAL events: 41
COMM events: 2 ( 4.9%)
EXIT events: 1 ( 2.4%)
SAMPLE events: 9 (22.0%)
MMAP2 events: 4 ( 9.8%)
ATTR events: 1 ( 2.4%)
FINISHED_ROUND events: 1 ( 2.4%)
THREAD_MAP events: 1 ( 2.4%)
CPU_MAP events: 1 ( 2.4%)
EVENT_UPDATE events: 1 ( 2.4%)
TIME_CONV events: 1 ( 2.4%)
FEATURE events: 19 (46.3%)
cycles:uhH stats:
SAMPLE events: 9
$
Fixes: a4a4d0a7a2b20f78 ("perf report: Add --stats option to display quick data statistics")
Signed-off-by: Namhyung Kim <namhyung(a)kernel.org>
Acked-by: Jiri Olsa <jolsa(a)redhat.com>
Tested-by: Arnaldo Carvalho de Melo <acme(a)redhat.com>
Cc: Andi Kleen <ak(a)linux.intel.com>
Cc: Ian Rogers <irogers(a)google.com>
Cc: Peter Zijlstra <peterz(a)infradead.org>
Link: http://lore.kernel.org/lkml/20210630043058.1131295-1-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme(a)redhat.com>
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index bc5c393021dc..8639bbe0969d 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -729,9 +729,14 @@ static int count_sample_event(struct perf_tool *tool __maybe_unused,
return 0;
}
+static int process_attr(struct perf_tool *tool __maybe_unused,
+ union perf_event *event,
+ struct evlist **pevlist);
+
static void stats_setup(struct report *rep)
{
memset(&rep->tool, 0, sizeof(rep->tool));
+ rep->tool.attr = process_attr;
rep->tool.sample = count_sample_event;
rep->tool.no_warn = true;
}
@@ -753,6 +758,7 @@ static void tasks_setup(struct report *rep)
rep->tool.mmap = perf_event__process_mmap;
rep->tool.mmap2 = perf_event__process_mmap2;
}
+ rep->tool.attr = process_attr;
rep->tool.comm = perf_event__process_comm;
rep->tool.exit = perf_event__process_exit;
rep->tool.fork = perf_event__process_fork;