As result, the metrics "LLC Ld Miss" is perfectly divided into two
metrics "RMT Load Hit" and "Load Dram". It's not necessary to keep
metrics "LLC Ld Miss", so remove it.
Leo Yan [Wed, 14 Oct 2020 05:09:20 +0000 (06:09 +0100)]
perf c2c: Correct LLC load hit metrics
"rmt_hit" is accounted into two metrics: one is accounted into the
metrics "LLC Ld Miss" (see the function llc_miss() for calculation
"llcmiss"); and it's accounted into metrics "LLC Load Hit". Thus,
for the literal meaning, it is contradictory that "rmt_hit" is
accounted for both "LLC Ld Miss" (LLC miss) and "LLC Load Hit"
(LLC hit).
Thus this is easily to introduce confusion: "LLC Load Hit" gives
impression that all items belong to it are LLC hit; in fact "rmt_hit"
is LLC miss and remote cache hit.
To give out clear semantics for metric "LLC Load Hit", "rmt_hit" is
moved out from it and changes "LLC Load Hit" to contain two items:
LLC Load Hit = LLC's hit ("ld_llchit") + LLC's hitm ("lcl_hitm")
For output alignment, adjusts the header for "LLC Load Hit".
Leo Yan [Wed, 14 Oct 2020 05:09:18 +0000 (06:09 +0100)]
perf c2c: Use more explicit headers for HITM
Local and remote HITM use the headers 'Lcl' and 'Rmt' respectively,
suppose if we want to extend the tool to display these two dimensions
under any one metrics, users cannot understand the semantics if only
based on the header string 'Lcl' or 'Rmt'.
To explicit express the meaning for HITM items, this patch changes the
headers string as "LclHitm" and "RmtHitm", the strings are more readable
and this allows to extend metrics for using HITM items.
Leo Yan [Wed, 14 Oct 2020 05:09:17 +0000 (06:09 +0100)]
perf c2c: Change header from "LLC Load Hitm" to "Load Hitm"
The metrics "LLC Load Hitm" contains two items: one is "local Hitm" and
another is "remote Hitm".
"local Hitm" means: L3 HIT and was serviced by another processor core
with a cross core snoop where modified copies were found; it's no doubt
that "local Hitm" belongs to LLC access.
But for "remote Hitm", based on the code in util/mem-events, it's the
event for remote cache HIT and was serviced by another processor core
with modified copies. Thus the remote Hitm is a remote cache's hit and
actually it's LLC load miss.
Now the display format gives users the impression that "local Hitm" and
"remote Hitm" both belong to the LLC load, but this is not the fact as
described.
This patch changes the header from "LLC Load Hitm" to "Load Hitm", this
can avoid the give the wrong impression that all Hitm belong to LLC.
Leo Yan [Wed, 14 Oct 2020 05:09:16 +0000 (06:09 +0100)]
perf c2c: Organize metrics based on memory hierarchy
The metrics are not organized based on memory hierarchy, e.g. the tool
doesn't organize the metrics order based on memory nodes from the close
node (e.g. L1/L2 cache) to far node (e.g. L3 cache and DRAM).
To output metrics with more friendly form, this patch refines the
metrics order based on memory hierarchy:
Leo Yan [Wed, 14 Oct 2020 05:09:15 +0000 (06:09 +0100)]
perf c2c: Display "Total Stores" as a standalone metrics
The total stores is displayed under the metrics "Store Reference", to
output the same format with total records and all loads, extract the
total stores number as a standalone metrics "Total Stores".
After this patch, the tool shows the summary numbers ("Total records",
"Total loads", "Total Stores") in the unified form.
Leo Yan [Wed, 14 Oct 2020 05:09:14 +0000 (06:09 +0100)]
perf c2c: Display the total numbers continuously
To view the statistics with "breakdown" mode, it's good to show the
summary numbers for the total records, all stores and all loads, then
the sequential conlumns can be used to break into more detailed items.
To achieve this purpose, this patch displays the summary numbers for
records/stores/loads continuously and places them before breakdown
items, this can allow uses to easily read the summarized statistics.
Ian Rogers [Mon, 12 Oct 2020 16:16:11 +0000 (09:16 -0700)]
perf bench: Use condition variables in numa.
The existing approach to synchronization between threads in the numa
benchmark is unbalanced mutexes.
This synchronization causes thread sanitizer to warn of locks being
taken twice on a thread without an unlock, as well as unlocks with no
corresponding locks.
This change replaces the synchronization with more regular condition
variables.
While this fixes one class of thread sanitizer warnings, there still
remain warnings of data races due to threads reading and writing shared
memory without any atomics.
Committer testing:
Basic run on a non-NUMA machine.
# perf bench numa
# List of available benchmarks for collection 'numa':
mem: Benchmark for NUMA workloads
all: Run all NUMA benchmarks
# perf bench numa all
# Running numa/mem benchmark...
# Running main, "perf bench numa numa-mem"
#
# Running test on: Linux five 5.8.12-200.fc32.x86_64 #1 SMP Mon Sep 28 12:17:31 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux
#
# Running RAM-bw-local, "perf bench numa mem -p 1 -t 1 -P 1024 -C 0 -M 0 -s 20 -zZq --thp 1 --no-data_rand_walk"
20.076 secs slowest (max) thread-runtime
20.000 secs fastest (min) thread-runtime
20.073 secs average thread-runtime
0.190 % difference between max/avg runtime
241.828 GB data processed, per thread
241.828 GB data processed, total
0.083 nsecs/byte/thread runtime
12.045 GB/sec/thread speed
12.045 GB/sec total speed
# Running RAM-bw-local-NOTHP, "perf bench numa mem -p 1 -t 1 -P 1024 -C 0 -M 0 -s 20 -zZq --thp 1 --no-data_rand_walk --thp -1"
20.045 secs slowest (max) thread-runtime
20.000 secs fastest (min) thread-runtime
20.014 secs average thread-runtime
0.111 % difference between max/avg runtime
234.304 GB data processed, per thread
234.304 GB data processed, total
0.086 nsecs/byte/thread runtime
11.689 GB/sec/thread speed
11.689 GB/sec total speed
Jin Yao [Fri, 9 Oct 2020 02:28:45 +0000 (10:28 +0800)]
perf diff: Support hot streams comparison
This patch enables perf-diff with "--stream" option.
"--stream": Enable hot streams comparison
Now let's see example.
perf record -b ... Generate perf.data.old with branch data
perf record -b ... Generate perf.data with branch data
perf diff --stream
[ Matched hot streams ]
hot chain pair 1:
cycles: 1, hits: 27.77% cycles: 1, hits: 9.24%
--------------------------- --------------------------
main div.c:39 main div.c:39
main div.c:44 main div.c:44
Jin Yao [Fri, 9 Oct 2020 02:28:41 +0000 (10:28 +0800)]
perf streams: Compare two streams
Stream is the branch history which is aggregated by the branch records
from perf samples. Now we support the callchain as stream.
If the callchain entries of one stream are fully matched with the
callchain entries of another stream, we think two streams are matched.
For example,
cycles: 1, hits: 26.80% cycles: 1, hits: 27.30%
----------------------- -----------------------
main div.c:39 main div.c:39
main div.c:44 main div.c:44
Above two streams are matched (we don't consider the case that source
code is changed).
The matching logic is, compare the chain string first. If it's not
matched, fallback to dso address comparison.
Jin Yao [Fri, 9 Oct 2020 02:28:39 +0000 (10:28 +0800)]
perf streams: Introduce branch history "streams"
We define a stream as the branch history which is aggregated by the
branch records from perf samples. For example, the callchains aggregated
from the branch records are considered as streams. By browsing the hot
stream, we can understand the hot code path.
Now we only support the callchain for stream. For measuring the hot
level for a stream, we use the callchain_node->hit, higher is hotter.
There may be many callchains sampled so we only focus on the top N
hottest callchains. N is a user defined parameter or predefined default
value (nr_streams_max).
This patch creates an evsel_streams array per event, and saves the top N
hottest streams in a stream array.
So now we can get the per-event top N hottest streams.
Andi Kleen [Wed, 14 Oct 2020 14:42:55 +0000 (07:42 -0700)]
perf tools: Add support for exclusive groups/events
Peter suggested that using the exclusive mode in perf could avoid some
problems with bad scheduling of groups. Exclusive is implemented in the
kernel, but wasn't exposed by the perf tool, so hard to use without
custom low level API users.
Add support for marking groups or events with :e for exclusive in the
perf tool. The implementation is basically the same as the existing
pinned attribute.
Committer testing:
# perf test "parse event"
6: Parse event definition strings : Ok
# perf test -v "parse event" |& grep :u*e
running test 56 'instructions:uep'
running test 57 '{cycles,cache-misses,branch-misses}:e'
#
#
# grep "model name" -m1 /proc/cpuinfo
model name : AMD Ryzen 9 3900X 12-Core Processor
#
# perf stat -a -e '{cycles,cache-misses,branch-misses}:e' sleep 1
#
# The output for asking for precise events on AMD needs to improve, it
# supposedly works only for system wide or per CPU
#
# perf stat -a -e '{cycles,cache-misses,branch-misses}:uep' sleep 1
Error:
The sys_perf_event_open() syscall returned with 22 (Invalid argument) for event (cycles).
/bin/dmesg | grep -i perf may provide additional information.
# perf stat -a -e '{cycles,cache-misses,branch-misses}:ue' sleep 1
Jiri Olsa [Tue, 13 Oct 2020 19:24:39 +0000 (21:24 +0200)]
perf tools: Add size to 'struct perf_record_header_build_id'
We do not store size with build ids in perf data, but there's enough
space to do it. Adding misc bit PERF_RECORD_MISC_BUILD_ID_SIZE to mark
build id event with size.
With this fix the dso with md5 build id will have correct build id data
and will be usable for debuginfod processing if needed (coming in
following patches).
Committer notes:
Use %zu with size_t to fix this error on 32-bit arches:
util/header.c: In function '__event_process_build_id':
util/header.c:2105:3: error: format '%lu' expects argument of type 'long unsigned int', but argument 6 has type 'size_t' [-Werror=format=]
pr_debug("build id event received for %s: %s [%lu]\n",
^
Arnaldo Carvalho de Melo [Tue, 13 Oct 2020 20:03:19 +0000 (17:03 -0300)]
perf config: Export the perf_config_from_file() function
We'll use it to ask for extra config files to be loaded, profile like
stuff that will be used first to make 'perf trace' mimic 'strace' output
via a 'perf strace' command that just sets up 'perf trace' output.
At some point it'll be used for regression tests, where we'll run some
simple commands like:
perf strace ls > perf-strace.output
strace ls > strace.output
And then do some mutable syscall arg aware diff like tool to deal with
arguments for things like mmap, that change at each execution, to be
first ignored and then properly tracked when used accoss multiple
syscalls.
Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
James Clark [Mon, 5 Oct 2020 08:06:45 +0000 (09:06 +0100)]
perf python: Autodetect python3 binary
Some distros don't come with python2 and only have python3 available.
This causes the "'import perf' in python" self test to fail.
This change adds python3 to the list of possible python versions
that are autodetected but maintains the priorities for
'python2' and 'python' detection. Python3 has the lowest priority.
Committer notes:
On a fedora system without python2 packages the 'perf test python'
continues to work:
# python2
bash: python2: command not found...
Similar command is: 'python'
# rpm -qa | grep python2
#
# rpm -e python-unversioned-command-3.8.5-5.fc32.noarch
# hash -r
# python
bash: python: command not found...
Install package 'python-unversioned-command' to provide command 'python'? [N/y] n
#
And rebuilding perf now doesn't find python in the system:
make: Entering directory '/home/acme/git/perf/tools/perf'
BUILD: Doing 'make -j24' parallel build
<SNIP>
Makefile.config:786: No python interpreter was found: disables Python support - please install python-devel/python-dev
<SNIP>
After this patch:
$ rpm -qi python-unversioned-command
package python-unversioned-command is not installed
$
$ python
bash: python: command not found...
Install package 'python-unversioned-command' to provide command 'python'? [N/y] ^C
$
$ m
make: Entering directory '/home/acme/git/perf/tools/perf'
BUILD: Doing 'make -j24' parallel build
<SNIP>
CC /tmp/build/perf/tests/attr.o
CC /tmp/build/perf/tests/python-use.o
DESCEND plugins
GEN /tmp/build/perf/python/perf.so
INSTALL trace_plugins
LD /tmp/build/perf/tests/perf-in.o
LD /tmp/build/perf/perf-in.o
LINK /tmp/build/perf/perf
<SNIP>
make: Leaving directory '/home/acme/git/perf/tools/perf'
19: 'import perf' in python : Ok
$ ldd ~/bin/perf | grep python
libpython3.8.so.1.0 => /lib64/libpython3.8.so.1.0 (0x00007f2c8c708000)
$ ls -la /tmp/build/perf/python
total 1864
drwxrwxr-x. 2 acme acme 60 Oct 13 16:20 .
drwxrwxr-x. 18 acme acme 4420 Oct 13 16:31 ..
-rwxrwxr-x. 1 acme acme 1907216 Oct 13 16:31 perf.cpython-38-x86_64-linux-gnu.so
$
Signed-off-by: James Clark <james.clark@arm.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org>
LPU-Reference: 20201005080645.6588-1-james.clark@arm.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Arnaldo Carvalho de Melo [Tue, 13 Oct 2020 19:22:03 +0000 (16:22 -0300)]
perf tests: Show python test script in verbose mode
To help figure out where it is getting the binding.
Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Jiri Slaby [Thu, 1 Oct 2020 09:34:19 +0000 (11:34 +0200)]
perf trace: Fix off by ones in memset() after realloc() in arches using libaudit
'perf trace ls' started crashing after commit d21cb73a9025 on
!HAVE_SYSCALL_TABLE_SUPPORT configs (armv7l here) like this:
0 strlen () at ../sysdeps/arm/armv6t2/strlen.S:126
1 0xb6800780 in __vfprintf_internal (s=0xbeff9908, s@entry=0xbeff9900, format=0xa27160 "]: %s()", ap=..., mode_flags=<optimized out>) at vfprintf-internal.c:1688
...
5 0x0056ecdc in fprintf (__fmt=0xa27160 "]: %s()", __stream=<optimized out>) at /usr/include/bits/stdio2.h:100
6 trace__sys_exit (trace=trace@entry=0xbeffc710, evsel=evsel@entry=0xd968d0, event=<optimized out>, sample=sample@entry=0xbeffc3e8) at builtin-trace.c:2475
7 0x00566d40 in trace__handle_event (sample=0xbeffc3e8, event=<optimized out>, trace=0xbeffc710) at builtin-trace.c:3122
...
15 main (argc=2, argv=0xbefff6e8) at perf.c:538
It is because memset in trace__read_syscall_info zeroes wrong memory:
1) when initializing for the first time, it does not reset the last id.
2) in other cases, it resets the last id of previous buffer.
ad 1) it causes the crash above as sc->name used in the fprintf above
contains garbage.
ad 2) it sets nonexistent from true back to false for id 11 here. Not
sure, what the consequences are.
So fix it by introducing a special case for the initial initialization
and do the right +1 in both cases.
Fixes: d21cb73a9025 ("perf trace: Grow the syscall table as needed when using libaudit") Signed-off-by: Jiri Slaby <jslaby@suse.cz> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lore.kernel.org/lkml/20201001093419.15761-1-jslaby@suse.cz Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Leo Yan [Sun, 11 Oct 2020 12:10:22 +0000 (20:10 +0800)]
perf c2c: Update usage for showing memory events
Since commit b027cc6fdf1b ("perf c2c: Fix 'perf c2c record -e list' to
show the default events used"), "perf c2c" tool can show the memory
events properly, it's no reason to still suggest user to use the
command "perf mem record -e list" for showing events.
This patch updates the usage for showing memory events with command
"perf c2c record -e list".
There are internal library functions, which are not declared as a static.
They are used inside the library from different files. Hide them from
the library users, as they are not part of the API.
These functions are made hidden and are renamed without the prefix "tep_":
tep_free_plugin_paths
tep_peek_char
tep_buffer_init
tep_get_input_buf_ptr
tep_get_input_buf
tep_read_token
tep_free_token
tep_free_event
tep_free_format_field
__tep_parse_format
The 'perf sched latency' tool is really useful at showing worst-case
latencies that task encountered since wakeup. However it shows only the
end of the latency. Often times the start of a latency is interesting as
it can show what else was going on at the time to cause the latency. I
certainly myself spending a lot of time backtracking to the start of the
latency in "perf sched script" which wastes a lot of time.
This patch therefore adds a new column "Max delay start". Considering
this, also rename "Maximum delay at" to "Max delay end" as its easier to
understand.
Example of the new output:
----------------------------------------------------------------------------------------------------------------------------------
Task | Runtime ms | Switches | Avg delay ms | Max delay ms | Max delay start | Max delay end |
----------------------------------------------------------------------------------------------------------------------------------
MediaScannerSer:11936 | 651.296 ms | 67978 | avg: 0.113 ms | max: 77.250 ms | max start: 477.691360 s | max end: 477.768610 s
audio@2.0-servi:(3) | 0.000 ms | 3440 | avg: 0.034 ms | max: 72.267 ms | max start: 477.697051 s | max end: 477.769318 s
AudioOut_1D:8112 | 0.000 ms | 2588 | avg: 0.083 ms | max: 64.020 ms | max start: 477.710740 s | max end: 477.774760 s
Time-limited te:14973 | 7966.090 ms | 24807 | avg: 0.073 ms | max: 15.563 ms | max start: 477.162746 s | max end: 477.178309 s
surfaceflinger:8049 | 9.680 ms | 603 | avg: 0.063 ms | max: 13.275 ms | max start: 476.931791 s | max end: 476.945067 s
HeapTaskDaemon:(3) | 1588.830 ms | 7040 | avg: 0.065 ms | max: 6.880 ms | max start: 473.666043 s | max end: 473.672922 s
mount-passthrou:(3) | 1370.809 ms | 68904 | avg: 0.011 ms | max: 6.524 ms | max start: 478.090630 s | max end: 478.097154 s
ReferenceQueueD:(3) | 11.794 ms | 1725 | avg: 0.014 ms | max: 6.521 ms | max start: 476.119782 s | max end: 476.126303 s
writer:14077 | 18.410 ms | 1427 | avg: 0.036 ms | max: 6.131 ms | max start: 474.169675 s | max end: 474.175805 s
Namhyung Kim [Mon, 12 Oct 2020 07:02:13 +0000 (16:02 +0900)]
perf inject: Add --buildid-all option
Like 'perf record', we can even more speedup build-id processing by just
using all DSOs. Then we don't need to look at all the sample events
anymore. The following patch will update 'perf bench' to show the result
of the --buildid-all option too.
Signed-off-by: Namhyung Kim <namhyung@kernel.org> Original-patch-by: Stephane Eranian <eranian@google.com> Acked-by: Ian Rogers <irogers@google.com> Acked-by: Jiri Olsa <jolsa@redhat.com> Link: https://lore.kernel.org/r/20201012070214.2074921-6-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Namhyung Kim [Mon, 12 Oct 2020 07:02:12 +0000 (16:02 +0900)]
perf inject: Do not load map/dso when injecting build-id
No need to load symbols in a DSO when injecting build-id. I guess the
reason was to check the DSO is a special file like anon files. Use some
helper functions in map.c to check them before reading build-id. Also
pass sample event's cpumode to a new build-id event.
It brought a speedup in the benchmark of 25 -> 21 msec on my laptop.
Also the memory usage (Max RSS) went down by ~200 KB.
# Running 'internals/inject-build-id' benchmark:
Average build-id injection took: 21.389 msec (+- 0.138 msec)
Average time per event: 2.097 usec (+- 0.014 usec)
Average memory usage: 8225 KB (+- 0 KB)
Committer notes:
Before:
$ perf stat -r5 perf bench internals inject-build-id > /dev/null
Performance counter stats for 'perf bench internals inject-build-id' (5 runs):
Namhyung Kim [Mon, 12 Oct 2020 07:02:11 +0000 (16:02 +0900)]
perf inject: Enter namespace when reading build-id
It should be in a proper mnt namespace when accessing the file.
I think this had no problem since the build-id was actually read from
map__load() -> dso__load() already. But I'd like to change it in the
following commit.
-i, --iterations <n> Number of iterations used to compute average (default: 100)
-m, --nr-mmaps <n> Number of mmap events for each iteration (default: 100)
-n, --nr-samples <n> Number of sample events per mmap event (default: 100)
-v, --verbose be more verbose (show iteration count, DSO name, etc)
By default, it measures average processing time of 100 MMAP2 events
and 10000 SAMPLE events. Below is a result on my laptop.
$ perf bench internals inject-build-id
# Running 'internals/inject-build-id' benchmark:
Average build-id injection took: 25.789 msec (+- 0.202 msec)
Average time per event: 2.528 usec (+- 0.020 usec)
Average memory usage: 8411 KB (+- 7 KB)
Namhyung Kim [Wed, 7 Oct 2020 08:13:11 +0000 (17:13 +0900)]
perf stat: Fix out of bounds CPU map access when handling armv8_pmu events
It was reported that 'perf stat' crashed when using with armv8_pmu (CPU)
events with the task mode. As 'perf stat' uses an empty cpu map for
task mode but armv8_pmu has its own cpu mask, it has confused which map
it should use when accessing file descriptors and this causes segfaults:
(gdb) bt
#0 0x0000000000603fc8 in perf_evsel__close_fd_cpu (evsel=<optimized out>,
cpu=<optimized out>) at evsel.c:122
#1 perf_evsel__close_cpu (evsel=evsel@entry=0x716e950, cpu=7) at evsel.c:156
#2 0x00000000004d4718 in evlist__close (evlist=0x70a7cb0) at util/evlist.c:1242
#3 0x0000000000453404 in __run_perf_stat (argc=3, argc@entry=1, argv=0x30,
argv@entry=0xfffffaea2f90, run_idx=119, run_idx@entry=1701998435)
at builtin-stat.c:929
#4 0x0000000000455058 in run_perf_stat (run_idx=1701998435, argv=0xfffffaea2f90,
argc=1) at builtin-stat.c:947
#5 cmd_stat (argc=1, argv=0xfffffaea2f90) at builtin-stat.c:2357
#6 0x00000000004bb888 in run_builtin (p=p@entry=0x9764b8 <commands+288>,
argc=argc@entry=4, argv=argv@entry=0xfffffaea2f90) at perf.c:312
#7 0x00000000004bbb54 in handle_internal_command (argc=argc@entry=4,
argv=argv@entry=0xfffffaea2f90) at perf.c:364
#8 0x0000000000435378 in run_argv (argcp=<synthetic pointer>,
argv=<synthetic pointer>) at perf.c:408
#9 main (argc=4, argv=0xfffffaea2f90) at perf.c:538
To fix this, I simply used the given cpu map unless the evsel actually
is not a system-wide event (like uncore events).
Fixes: 7736627b865d ("perf stat: Use affinity for closing file descriptors") Reported-by: Wei Li <liwei391@huawei.com> Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Barry Song <song.bao.hua@hisilicon.com> Acked-by: Jiri Olsa <jolsa@redhat.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lore.kernel.org/lkml/20201007081311.1831003-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The problem is in the is_printable_array function that does not take the
zero byte into account and claim such string as not printable, so the
code will create byte array instead of string.
Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
It'll also conditionally generate the defines, so that if we don't have
those when building a new tool tarball in an older systems, we get
those, and we need them sometimes in the actual scnprintf routine, such
as when checking if a flags means we have an extra arg, like with
MREMAP_FIXED.
Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf tools: Separate the checking of headers only used to build beautification tables
Some headers are not used in building the tools directly, but instead to
generate tables that then gets source code included to do id->string and
string->id lookups for things like syscall flags and commands.
We were adding it directly to tools/include/ and this sometimes gets in
the way of building using system headers, lets untangle this a bit.
Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Merge tag 'nfs-for-5.9-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Pull NFS client bugfixes from Trond Myklebust:
"Highlights include:
- NFSv4.2: copy_file_range needs to invalidate caches on success
- NFSv4.2: Fix security label length not being reset
- pNFS/flexfiles: Ensure we initialise the mirror bsizes correctly
on read
- pNFS/flexfiles: Fix signed/unsigned type issues with mirror
indices"
* tag 'nfs-for-5.9-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
pNFS/flexfiles: Be consistent about mirror index types
pNFS/flexfiles: Ensure we initialise the mirror bsizes correctly on read
NFSv4.2: fix client's attribute cache management for copy_file_range
nfs: Fix security label length not being reset
Jason A. Donenfeld [Mon, 28 Sep 2020 10:35:07 +0000 (12:35 +0200)]
mm: do not rely on mm == current->mm in __get_user_pages_locked
It seems likely this block was pasted from internal_get_user_pages_fast,
which is not passed an mm struct and therefore uses current's. But
__get_user_pages_locked is passed an explicit mm, and current->mm is not
always valid. This was hit when being called from i915, which uses:
This commit fixes the problem by using the mm pointer passed to the
function rather than the bogus one in current.
Fixes: 008cfe4418b3 ("mm: Introduce mm_struct.has_pinned") Tested-by: Chris Wilson <chris@chris-wilson.co.uk> Reported-by: Harald Arnesen <harald@skogtun.org> Reviewed-by: Jason Gunthorpe <jgg@nvidia.com> Reviewed-by: Peter Xu <peterx@redhat.com> Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Ian Rogers [Wed, 23 Sep 2020 21:06:55 +0000 (14:06 -0700)]
perf test: Fix msan uninitialized use.
Ensure 'st' is initialized before an error branch is taken.
Fixes test "67: Parse and process metrics" with LLVM msan:
==6757==WARNING: MemorySanitizer: use-of-uninitialized-value
#0 0x5570edae947d in rblist__exit tools/perf/util/rblist.c:114:2
#1 0x5570edb1c6e8 in runtime_stat__exit tools/perf/util/stat-shadow.c:141:2
#2 0x5570ed92cfae in __compute_metric tools/perf/tests/parse-metric.c:187:2
#3 0x5570ed92cb74 in compute_metric tools/perf/tests/parse-metric.c:196:9
#4 0x5570ed92c6d8 in test_recursion_fail tools/perf/tests/parse-metric.c:318:2
#5 0x5570ed92b8c8 in test__parse_metric tools/perf/tests/parse-metric.c:356:2
#6 0x5570ed8de8c1 in run_test tools/perf/tests/builtin-test.c:410:9
#7 0x5570ed8ddadf in test_and_print tools/perf/tests/builtin-test.c:440:9
#8 0x5570ed8dca04 in __cmd_test tools/perf/tests/builtin-test.c:661:4
#9 0x5570ed8dbc07 in cmd_test tools/perf/tests/builtin-test.c:807:9
#10 0x5570ed7326cc in run_builtin tools/perf/perf.c:313:11
#11 0x5570ed731639 in handle_internal_command tools/perf/perf.c:365:8
#12 0x5570ed7323cd in run_argv tools/perf/perf.c:409:2
#13 0x5570ed731076 in main tools/perf/perf.c:539:3
Fixes: commit f5a56570a3f2 ("perf test: Fix memory leaks in parse-metric test") Signed-off-by: Ian Rogers <irogers@google.com> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: clang-built-linux@googlegroups.com Link: http://lore.kernel.org/lkml/20200923210655.4143682-1-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Ian Rogers [Fri, 25 Sep 2020 00:39:03 +0000 (17:39 -0700)]
perf parse-events: Reduce casts around bp_addr
perf_event_attr bp_addr is a u64. parse-events.y parses it as a u64, but
casts it to a void* and then parse-events.c casts it back to a u64.
Rather than all the casts, change the type of the address to be a u64.
Signed-off-by: Ian Rogers <irogers@google.com> Acked-by: Jiri Olsa <jolsa@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Jin Yao <yao.jin@linux.intel.com> Cc: Leo Yan <leo.yan@linaro.org> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lore.kernel.org/lkml/20200925003903.561568-1-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Namhyung Kim [Thu, 24 Sep 2020 12:44:54 +0000 (21:44 +0900)]
perf tools: Allow creation of cgroup without open
This is a preparation for a test case of expanding events for multiple
cgroups. Instead of using real system cgroup, the test will use fake
cgroups so it needs a way to have them without a open file descriptor.
Signed-off-by: Namhyung Kim <namhyung@kernel.org> Acked-by: Jiri Olsa <jolsa@redhat.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lore.kernel.org/lkml/20200924124455.336326-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Namhyung Kim [Thu, 24 Sep 2020 12:44:53 +0000 (21:44 +0900)]
perf tools: Copy metric events properly when expand cgroups
The metricgroup__copy_metric_events() is to handle metrics events when
expanding event for cgroups. As the metric events keep pointers to
evsel, it should be refreshed when events are cloned during the
operation.
The perf_stat__collect_metric_expr() is also called in case an event has
a metric directly.
During the copy, it references evsel by index as the evlist now has
cloned evsels for the given cgroup.
Also kernel test robot found an issue in the python module import so add
empty implementations of those two functions to fix it.
Reported-by: kernel test robot <rong.a.chen@intel.com> Signed-off-by: Namhyung Kim <namhyung@kernel.org> Acked-by: Jiri Olsa <jolsa@redhat.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Ian Rogers <irogers@google.com> Cc: John Garry <john.garry@huawei.com> Cc: Kajol Jain <kjain@linux.ibm.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lore.kernel.org/lkml/20200924124455.336326-4-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Namhyung Kim [Thu, 24 Sep 2020 12:44:52 +0000 (21:44 +0900)]
perf stat: Add --for-each-cgroup option
The --for-each-cgroup option is a syntax sugar to monitor large number
of cgroups easily. Current command line requires to list all the events
and cgroups even if users want to monitor same events for each cgroup.
This patch addresses that usage by copying given events for each cgroup
on user's behalf.
For instance, if they want to monitor 6 events for 200 cgroups each they
should write 1200 event names (with -e) AND 1200 cgroup names (with -G)
on the command line. But with this change, they can just specify 6
events and 200 cgroups with a new option.
A simpler example below: It wants to measure 3 events for 2 cgroups ('A'
and 'B'). The result is that total 6 events are counted like below.
$ perf stat -a -e cpu-clock,cycles,instructions --for-each-cgroup A,B sleep 1
Performance counter stats for 'system wide':
988.18 msec cpu-clock A # 0.987 CPUs utilized
3,153,761,702 cycles A # 3.200 GHz (100.00%)
8,067,769,847 instructions A # 2.57 insn per cycle (100.00%)
982.71 msec cpu-clock B # 0.982 CPUs utilized
3,136,093,298 cycles B # 3.182 GHz (99.99%)
8,109,619,327 instructions B # 2.58 insn per cycle (99.99%)
Namhyung Kim [Thu, 24 Sep 2020 12:44:51 +0000 (21:44 +0900)]
perf evsel: Add evsel__clone() function
The evsel__clone() is to create an exactly same evsel from same
attributes. The function assumes the given evsel is not configured
yet so it cares fields set during event parsing. Those fields are now
moved together as Jiri suggested. Note that metric events will be
handled by later patch.
It will be used by perf stat to generate separate events for each
cgroup.
Signed-off-by: Namhyung Kim <namhyung@kernel.org> Acked-by: Jiri Olsa <jolsa@redhat.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lore.kernel.org/lkml/20200924124455.336326-2-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Jin Yao <yao.jin@linux.intel.com> Reviewed-by: Andi Kleen <ak@linux.intel.com> Acked-by: Ian Rogers <irogers@google.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/lkml/20200922031918.3723-1-yao.jin@linux.intel.com/ Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Jin Yao [Tue, 22 Sep 2020 02:51:19 +0000 (10:51 +0800)]
perf vendor events intel: Update CascadelakeX events to v1.08
- Update CascadelakeX events to v1.08.
- Update CascadelakeX JSON metrics from TMAM 4.0.
Other fixes:
- Add NO_NMI_WATCHDOG metric constraint to Backend_Bound
- Change 'MB/sec' to 'MB' in UNC_M_PMM_BANDWIDTH.
Signed-off-by: Jin Yao <yao.jin@linux.intel.com> Reviewed-by: Andi Kleen <ak@linux.intel.com> Acked-by: Ian Rogers <irogers@google.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Kan Liang <kan.liang@intel.com> Link: https://lore.kernel.org/lkml/20200922031918.3723-1-yao.jin@linux.intel.com/ Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Merge tag 'x86-urgent-2020-09-27' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 fixes from Thomas Gleixner:
"Two fixes for the x86 interrupt code:
- Unbreak the magic 'search the timer interrupt' logic in IO/APIC
code which got wreckaged when the core interrupt code made the
state tracking logic stricter.
That caused the interrupt line to stay masked after switching from
IO/APIC to PIC delivery mode, which obviously prevents interrupts
from being delivered.
- Make run_on_irqstack_code() typesafe. The function argument is a
void pointer which is then cast to 'void (*fun)(void *).
This breaks Control Flow Integrity checking in clang. Use proper
helper functions for the three variants reuqired"
* tag 'x86-urgent-2020-09-27' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/ioapic: Unbreak check_timer()
x86/irq: Make run_on_irqstack_cond() typesafe
Merge tag 'timers-urgent-2020-09-27' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull timer updates from Thomas Gleixner:
"A set of clocksource/clockevents updates:
- Reset the TI/DM timer before enabling it instead of doing it the
other way round.
- Initialize the reload value for the GX6605s timer correctly so the
hardware counter starts at 0 again after overrun.
- Make error return value negative in the h8300 timer init function"
* tag 'timers-urgent-2020-09-27' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
clocksource/drivers/timer-gx6605s: Fixup counter reload
clocksource/drivers/timer-ti-dm: Do reset before enable
clocksource/drivers/h8300_timer8: Fix wrong return value in h8300_8timer_init()
Peter Xu [Fri, 25 Sep 2020 22:26:00 +0000 (18:26 -0400)]
mm/thp: Split huge pmds/puds if they're pinned when fork()
Pinned pages shouldn't be write-protected when fork() happens, because
follow up copy-on-write on these pages could cause the pinned pages to
be replaced by random newly allocated pages.
For huge PMDs, we split the huge pmd if pinning is detected. So that
future handling will be done by the PTE level (with our latest changes,
each of the small pages will be copied). We can achieve this by let
copy_huge_pmd() return -EAGAIN for pinned pages, so that we'll
fallthrough in copy_pmd_range() and finally land the next
copy_pte_range() call.
Huge PUDs will be even more special - so far it does not support
anonymous pages. But it can actually be done the same as the huge PMDs
even if the split huge PUDs means to erase the PUD entries. It'll
guarantee the follow up fault ins will remap the same pages in either
parent/child later.
This might not be the most efficient way, but it should be easy and
clean enough. It should be fine, since we're tackling with a very rare
case just to make sure userspaces that pinned some thps will still work
even without MADV_DONTFORK and after they fork()ed.
Signed-off-by: Peter Xu <peterx@redhat.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Peter Xu [Fri, 25 Sep 2020 22:25:59 +0000 (18:25 -0400)]
mm: Do early cow for pinned pages during fork() for ptes
This allows copy_pte_range() to do early cow if the pages were pinned on
the source mm.
Currently we don't have an accurate way to know whether a page is pinned
or not. The only thing we have is page_maybe_dma_pinned(). However
that's good enough for now. Especially, with the newly added
mm->has_pinned flag to make sure we won't affect processes that never
pinned any pages.
It would be easier if we can do GFP_KERNEL allocation within
copy_one_pte(). Unluckily, we can't because we're with the page table
locks held for both the parent and child processes. So the page
allocation needs to be done outside copy_one_pte().
Some trick is there in copy_present_pte(), majorly the wrprotect trick
to block concurrent fast-gup. Comments in the function should explain
better in place.
Oleg Nesterov reported a (probably harmless) bug during review that we
didn't reset entry.val properly in copy_pte_range() so that potentially
there's chance to call add_swap_count_continuation() multiple times on
the same swp entry. However that should be harmless since even if it
happens, the same function (add_swap_count_continuation()) will return
directly noticing that there're enough space for the swp counter. So
instead of a standalone stable patch, it is touched up in this patch
directly.
Peter Xu [Fri, 25 Sep 2020 22:25:57 +0000 (18:25 -0400)]
mm: Introduce mm_struct.has_pinned
(Commit message majorly collected from Jason Gunthorpe)
Reduce the chance of false positive from page_maybe_dma_pinned() by
keeping track if the mm_struct has ever been used with pin_user_pages().
This allows cases that might drive up the page ref_count to avoid any
penalty from handling dma_pinned pages.
Future work is planned, to provide a more sophisticated solution, likely
to turn it into a real counter. For now, make it atomic_t but use it as
a boolean for simplicity.
Suggested-by: Jason Gunthorpe <jgg@ziepe.ca> Signed-off-by: Peter Xu <peterx@redhat.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Merge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Pull SCSI fixes from James Bottomley:
"Three fixes: one in drivers (lpfc) and two for zoned block devices.
The latter also impinges on the block layer but only to introduce a
new block API for setting the zone model rather than fiddling with the
queue directly in the zoned block driver"
* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
scsi: sd: sd_zbc: Fix ZBC disk initialization
scsi: sd: sd_zbc: Fix handling of host-aware ZBC disks
scsi: lpfc: Fix initial FLOGI failure due to BBSCN not supported
Merge tag 'io_uring-5.9-2020-09-25' of git://git.kernel.dk/linux-block
Pull io_uring fixes from Jens Axboe:
"Two fixes for regressions in this cycle, and one that goes to 5.8
stable:
- fix leak of getname() retrieved filename
- remove plug->nowait assignment, fixing a regression with btrfs
- fix for async buffered retry"
* tag 'io_uring-5.9-2020-09-25' of git://git.kernel.dk/linux-block:
io_uring: ensure async buffered read-retry is setup properly
io_uring: don't unconditionally set plug->nowait = true
io_uring: ensure open/openat2 name is cleaned on cancelation
Merge tag 'block-5.9-2020-09-25' of git://git.kernel.dk/linux-block
Pull block fixes from Jens Axboe:
"NVMe pull request from Christoph, and removal of a dead define.
- fix error during controller probe that cause double free irqs
(Keith Busch)
- FC connection establishment fix (James Smart)
- properly handle completions for invalid tags (Xianting Tian)
- pass the correct nsid to the command effects and supported log
(Chaitanya Kulkarni)"
* tag 'block-5.9-2020-09-25' of git://git.kernel.dk/linux-block:
block: remove unused BLK_QC_T_EAGAIN flag
nvme-core: don't use NVME_NSID_ALL for command effects and supported log
nvme-fc: fail new connections to a deleted host or remote port
nvme-pci: fix NULL req in completion handler
nvme: return errors for hwmon init
Subsystems affected by this patch series: mm (thp, memcg, gup,
migration, memory-hotplug), lib, and x86"
* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
mm: don't rely on system state to detect hot-plug operations
mm: replace memmap_context by meminit_context
arch/x86/lib/usercopy_64.c: fix __copy_user_flushcache() cache writeback
lib/memregion.c: include memregion.h
lib/string.c: implement stpcpy
mm/migrate: correct thp migration stats
mm/gup: fix gup_fast with dynamic page table folding
mm: memcontrol: fix missing suffix of workingset_restore
mm, THP, swap: fix allocating cluster for swapfile by mistake
mm: don't rely on system state to detect hot-plug operations
In register_mem_sect_under_node() the system_state's value is checked to
detect whether the call is made during boot time or during an hot-plug
operation. Unfortunately, that check against SYSTEM_BOOTING is wrong
because regular memory is registered at SYSTEM_SCHEDULING state. In
addition, memory hot-plug operation can be triggered at this system
state by the ACPI [1]. So checking against the system state is not
enough.
The consequence is that on system with interleaved node's ranges like this:
This can be seen on PowerPC LPAR after multiple memory hot-plug and
hot-unplug operations are done. At the next reboot the node's memory
ranges can be interleaved and since the call to link_mem_sections() is
made in topology_init() while the system is in the SYSTEM_SCHEDULING
state, the node's id is not checked, and the sections registered to
multiple nodes:
$ ls -l /sys/devices/system/memory/memory21/node*
total 0
lrwxrwxrwx 1 root root 0 Aug 24 05:27 node1 -> ../../node/node1
lrwxrwxrwx 1 root root 0 Aug 24 05:27 node2 -> ../../node/node2
In that case, the system is able to boot but if later one of theses
memory blocks is hot-unplugged and then hot-plugged, the sysfs
inconsistency is detected and this is triggering a BUG_ON():
This patch addresses the root cause by not relying on the system_state
value to detect whether the call is due to a hot-plug operation. An
extra parameter is added to link_mem_sections() detailing whether the
operation is due to a hot-plug operation.
[1] According to Oscar Salvador, using this qemu command line, ACPI
memory hotplug operations are raised at SYSTEM_SCHEDULING state:
In that case, we can see memory blocks assigned to multiple nodes in
sysfs:
$ ls -l /sys/devices/system/memory/memory21
total 0
lrwxrwxrwx 1 root root 0 Aug 24 05:27 node1 -> ../../node/node1
lrwxrwxrwx 1 root root 0 Aug 24 05:27 node2 -> ../../node/node2
-rw-r--r-- 1 root root 65536 Aug 24 05:27 online
-r--r--r-- 1 root root 65536 Aug 24 05:27 phys_device
-r--r--r-- 1 root root 65536 Aug 24 05:27 phys_index
drwxr-xr-x 2 root root 0 Aug 24 05:27 power
-r--r--r-- 1 root root 65536 Aug 24 05:27 removable
-rw-r--r-- 1 root root 65536 Aug 24 05:27 state
lrwxrwxrwx 1 root root 0 Aug 24 05:25 subsystem -> ../../../../bus/memory
-rw-r--r-- 1 root root 65536 Aug 24 05:25 uevent
-r--r--r-- 1 root root 65536 Aug 24 05:27 valid_zones
The same applies in the node's directory with a memory21 link in both
the node1 and node2's directory.
This is wrong but doesn't prevent the system to run. However when
later, one of these memory blocks is hot-unplugged and then hot-plugged,
the system is detecting an inconsistency in the sysfs layout and a
BUG_ON() is raised:
The root cause of this issue is that when node's memory is registered,
the range used can overlap another node's range, thus the memory block
is registered to multiple nodes in sysfs.
There are two issues here:
(a) The sysfs memory and node's layouts are broken due to these
multiple links
(b) The link errors in link_mem_sections() should not lead to a system
panic.
To address (a) register_mem_sect_under_node should not rely on the
system state to detect whether the link operation is triggered by a hot
plug operation or not. This is addressed by the patches 1 and 2 of this
series.
Issue (b) will be addressed separately.
This patch (of 2):
The memmap_context enum is used to detect whether a memory operation is
due to a hot-add operation or happening at boot time.
Make it general to the hotplug operation and rename it as
meminit_context.
There is no functional change introduced by this patch
Suggested-by: David Hildenbrand <david@redhat.com> Signed-off-by: Laurent Dufour <ldufour@linux.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: David Hildenbrand <david@redhat.com> Reviewed-by: Oscar Salvador <osalvador@suse.de> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Rafael J . Wysocki" <rafael@kernel.org> Cc: Nathan Lynch <nathanl@linux.ibm.com> Cc: Scott Cheloha <cheloha@linux.ibm.com> Cc: Tony Luck <tony.luck@intel.com> Cc: Fenghua Yu <fenghua.yu@intel.com> Cc: <stable@vger.kernel.org> Link: https://lkml.kernel.org/r/20200915094143.79181-1-ldufour@linux.ibm.com Link: https://lkml.kernel.org/r/20200915132624.9723-1-ldufour@linux.ibm.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Jason Yan [Sat, 26 Sep 2020 04:19:21 +0000 (21:19 -0700)]
lib/memregion.c: include memregion.h
This addresses the following sparse warning:
lib/memregion.c:8:5: warning: symbol 'memregion_alloc' was not declared. Should it be static?
lib/memregion.c:14:6: warning: symbol 'memregion_free' was not declared. Should it be static?
Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Jason Yan <yanaijie@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Link: https://lkml.kernel.org/r/20200921142852.875312-1-yanaijie@huawei.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Nick Desaulniers [Sat, 26 Sep 2020 04:19:18 +0000 (21:19 -0700)]
lib/string.c: implement stpcpy
LLVM implemented a recent "libcall optimization" that lowers calls to
`sprintf(dest, "%s", str)` where the return value is used to
`stpcpy(dest, str) - dest`.
This generally avoids the machinery involved in parsing format strings.
`stpcpy` is just like `strcpy` except it returns the pointer to the new
tail of `dest`. This optimization was introduced into clang-12.
Implement this so that we don't observe linkage failures due to missing
symbol definitions for `stpcpy`.
Similar to last year's fire drill with: commit 5f074f3e192f
("lib/string.c: implement a basic bcmp")
The kernel is somewhere between a "freestanding" environment (no full
libc) and "hosted" environment (many symbols from libc exist with the
same type, function signature, and semantics).
As Peter Anvin notes, there's not really a great way to inform the
compiler that you're targeting a freestanding environment but would like
to opt-in to some libcall optimizations (see pr/47280 below), rather
than opt-out.
Arvind notes, -fno-builtin-* behaves slightly differently between GCC
and Clang, and Clang is missing many __builtin_* definitions, which I
consider a bug in Clang and am working on fixing.
Masahiro summarizes the subtle distinction between compilers justly:
To prevent transformation from foo() into bar(), there are two ways in
Clang to do that; -fno-builtin-foo, and -fno-builtin-bar. There is
only one in GCC; -fno-buitin-foo.
(Any difference in that behavior in Clang is likely a bug from a missing
__builtin_* definition.)
Masahiro also notes:
We want to disable optimization from foo() to bar(),
but we may still benefit from the optimization from
foo() into something else. If GCC implements the same transform, we
would run into a problem because it is not -fno-builtin-bar, but
-fno-builtin-foo that disables that optimization.
In this regard, -fno-builtin-foo would be more future-proof than
-fno-built-bar, but -fno-builtin-foo is still potentially overkill. We
may want to prevent calls from foo() being optimized into calls to
bar(), but we still may want other optimization on calls to foo().
It seems that compilers today don't quite provide the fine grain control
over which libcall optimizations pseudo-freestanding environments would
prefer.
Finally, Kees notes that this interface is unsafe, so we should not
encourage its use. As such, I've removed the declaration from any
header, but it still needs to be exported to avoid linkage errors in
modules.
Zi Yan [Sat, 26 Sep 2020 04:19:14 +0000 (21:19 -0700)]
mm/migrate: correct thp migration stats
PageTransHuge returns true for both thp and hugetlb, so thp stats was
counting both thp and hugetlb migrations. Exclude hugetlb migration by
setting is_thp variable right.
Clean up thp handling code too when we are there.
Fixes: 1a5bae25e3cf ("mm/vmstat: add events for THP migration without split") Signed-off-by: Zi Yan <ziy@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Link: https://lkml.kernel.org/r/20200917210413.1462975-1-zi.yan@sent.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/gup: fix gup_fast with dynamic page table folding
Currently to make sure that every page table entry is read just once
gup_fast walks perform READ_ONCE and pass pXd value down to the next
gup_pXd_range function by value e.g.:
static int gup_pud_range(p4d_t p4d, unsigned long addr, unsigned long end,
unsigned int flags, struct page **pages, int *nr)
...
pudp = pud_offset(&p4d, addr);
This function passes a reference on that local value copy to pXd_offset,
and might get the very same pointer in return. This happens when the
level is folded (on most arches), and that pointer should not be
iterated.
On s390 due to the fact that each task might have different 5,4 or
3-level address translation and hence different levels folded the logic
is more complex and non-iteratable pointer to a local copy leads to
severe problems.
Here is an example of what happens with gup_fast on s390, for a task
with 3-level paging, crossing a 2 GB pud boundary:
// addr = 0x1007ffff000, end = 0x10080001000
static int gup_pud_range(p4d_t p4d, unsigned long addr, unsigned long end,
unsigned int flags, struct page **pages, int *nr)
{
unsigned long next;
pud_t *pudp;
// pud_offset returns &p4d itself (a pointer to a value on stack)
pudp = pud_offset(&p4d, addr);
do {
// on second iteratation reading "random" stack value
pud_t pud = READ_ONCE(*pudp);
// next = 0x10080000000, due to PUD_SIZE/MASK != PGDIR_SIZE/MASK on s390
next = pud_addr_end(addr, end);
...
} while (pudp++, addr = next, addr != end); // pudp++ iterating over stack
return 1;
}
This happens since s390 moved to common gup code with commit d1874a0c2805 ("s390/mm: make the pxd_offset functions more robust") and
commit 1a42010cdc26 ("s390/mm: convert to the generic
get_user_pages_fast code").
s390 tried to mimic static level folding by changing pXd_offset
primitives to always calculate top level page table offset in pgd_offset
and just return the value passed when pXd_offset has to act as folded.
What is crucial for gup_fast and what has been overlooked is that
PxD_SIZE/MASK and thus pXd_addr_end should also change correspondingly.
And the latter is not possible with dynamic folding.
To fix the issue in addition to pXd values pass original pXdp pointers
down to gup_pXd_range functions. And introduce pXd_offset_lockless
helpers, which take an additional pXd entry value parameter. This has
already been discussed in
mm, THP, swap: fix allocating cluster for swapfile by mistake
SWP_FS is used to make swap_{read,write}page() go through the
filesystem, and it's only used for swap files over NFS. So, !SWP_FS
means non NFS for now, it could be either file backed or device backed.
Something similar goes with legacy SWP_FILE.
So in order to achieve the goal of the original patch, SWP_BLKDEV should
be used instead.
FS corruption can be observed with SSD device + XFS + fragmented
swapfile due to CONFIG_THP_SWAP=y.
I reproduced the issue with the following details:
mm: slab: fix potential double free in ___cache_free
With the commit 10befea91b61 ("mm: memcg/slab: use a single set of
kmem_caches for all allocations"), it becomes possible to call kfree()
from the slabs_destroy().
The functions cache_flusharray() and do_drain() calls slabs_destroy() on
array_cache of the local CPU without updating the size of the
array_cache. This enables the kfree() call from the slabs_destroy() to
recursively call cache_flusharray() which can potentially call
free_block() on the same elements of the array_cache of the local CPU
and causing double free and memory corruption.
To fix the issue, simply update the local CPU array_cache cache before
calling slabs_destroy().
Fixes: 10befea91b61 ("mm: memcg/slab: use a single set of kmem_caches for all allocations") Signed-off-by: Shakeel Butt <shakeelb@google.com> Reviewed-by: Roman Gushchin <guro@fb.com> Tested-by: Ming Lei <ming.lei@redhat.com> Reported-by: kernel test robot <rong.a.chen@intel.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Ted Ts'o <tytso@mit.edu> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Pull more kvm fixes from Paolo Bonzini:
"Five small fixes.
The nested migration bug will be fixed with a better API in 5.10 or
5.11, for now this is a fix that works with existing userspace but
keeps the current ugly API"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
KVM: SVM: Add a dedicated INVD intercept routine
KVM: x86: Reset MMU context if guest toggles CR4.SMAP or CR4.PKE
KVM: x86: fix MSR_IA32_TSC read for nested migration
selftests: kvm: Fix assert failure in single-step test
KVM: x86: VMX: Make smaller physical guest address space support user-configurable
Merge tag 'mips_fixes_5.9_3' of git://git.kernel.org/pub/scm/linux/kernel/git/mips/linux
Pull MIPS fixes from Thomas Bogendoerfer:
- fixed FP register access on Loongsoon-3
- added missing 1074 cpu handling
- fixed Loongson2ef build error
* tag 'mips_fixes_5.9_3' of git://git.kernel.org/pub/scm/linux/kernel/git/mips/linux:
MIPS: BCM47XX: Remove the needless check with the 1074K
MIPS: Add the missing 'CPU_1074K' into __get_cpu_type()
MIPS: Loongson2ef: Disable Loongson MMI instructions
MIPS: Loongson-3: Fix fp register access if MSA enabled
Merge tag 'spi-fix-v5.9-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi
Pull spi fixes from Mark Brown:
"A small collection of driver specific fixes, the fsl-espi and bcm-qspi
changes in particular have been causing breakage for users"
* tag 'spi-fix-v5.9-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi:
spi: bcm-qspi: Fix probe regression on iProc platforms
spi: fsl-dspi: fix use-after-free in remove path
spi: fsl-espi: Only process interrupts for expected events
spi: bcm2835: Make polling_limit_us static
spi: spi-fsl-dspi: use XSPI mode instead of DMA for DPAA2 SoCs
Merge tag 'regulator-fix-v5.9-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator
Pull regulator fix from Mark Brown:
"A single fix for incorrect specification of some of the register
fields on axp20x devices which would break voltage setting on affected
systems"
* tag 'regulator-fix-v5.9-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator:
regulator: axp20x: fix LDO2/4 description
Merge tag 'regmap-fix-v5.9-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap
Pull regmap fixes from Mark Brown:
"Two issues here - one is a fix for use after free issues in the case
where a regmap overrides its name using something dynamically
generated, the other is that we weren't handling access checks
non-incrementing I/O on registers within paged register regions
correctly resulting in spurious errors.
Both of these are quite rare but serious if they occur"
* tag 'regmap-fix-v5.9-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap:
regmap: fix page selection for noinc writes
regmap: fix page selection for noinc reads
regmap: debugfs: Add back in erroneously removed initialisation of ret
regmap: debugfs: Fix handling of name string for debugfs init delays
io_uring: ensure async buffered read-retry is setup properly
A previous commit for fixing up short reads botched the async retry
path, so we ended up going to worker threads more often than we should.
Fix this up, so retries work the way they originally were intended to.
Merge tag 'pm-5.9-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull power management fixes from Rafael Wysocki:
"These fix more fallout of recent RCU-lockdep changes in CPU idle code
and two devfreq issues.
Specifics:
- Export rcu_idle_{enter,exit} to modules to fix build issues
introduced by recent RCU-lockdep fixes (Borislav Petkov)
- Add missing return statement to a stub function in the ACPI
processor driver to fix a build issue introduced by recent
RCU-lockdep fixes (Rafael Wysocki)
- Fix recently introduced suspicious RCU usage warnings in the PSCI
cpuidle driver and drop stale comments regarding RCU_NONIDLE()
usage from enter_s2idle_proper() (Ulf Hansson)
- Fix error code path in the tegra30 devfreq driver (Dan Carpenter)
- Add missing information to devfreq_summary debugfs (Chanwoo Choi)"
* tag 'pm-5.9-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
ACPI: processor: Fix build for ARCH_APICTIMER_STOPS_ON_C3 unset
PM / devfreq: tegra30: Disable clock on error in probe
PM / devfreq: Add timer type to devfreq_summary debugfs
cpuidle: Drop misleading comments about RCU usage
cpuidle: psci: Fix suspicious RCU usage
rcu/tree: Export rcu_idle_{enter,exit} to modules
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma
Pull rdma fix from Jason Gunthorpe:
"One fix for a bug that blktests hits when using rxe: tear down the CQ
pool before waiting for all references to go away"
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma:
RDMA/core: Fix ordering of CQ pool destruction
Merge tag 'drm-fixes-2020-09-25' of git://anongit.freedesktop.org/drm/drm
Pull drm fixes from Dave Airlie:
"Fairly quiet, a couple of i915 fixes, one dma-buf fix, one vc4 and two
sun4i changes
dma-buf:
- Single null pointer deref fix
i915:
- Fix selftest reference to stack data out of scope
- Fix GVT null pointer dereference
vc4:
- fill asoc card owner
sun4i:
- program secondary CSC correctly"
* tag 'drm-fixes-2020-09-25' of git://anongit.freedesktop.org/drm/drm:
drm/i915/selftests: Push the fake iommu device from the stack to data
dmabuf: fix NULL pointer dereference in dma_buf_release()
drm/i915/gvt: Fix port number for BDW on EDID region setup
drm/sun4i: mixer: Extend regmap max_register
drm/sun4i: sun8i-csc: Secondary CSC register correction
drm/vc4/vc4_hdmi: fill ASoC card owner