Commit Graph

693 Commits

Author SHA1 Message Date
Mike Hommey
d1c271f02d Bug 1321093 - Avoid logalloc-replay make check failure when libstdc++ allocates memory in some static initializer. r=njn
--HG--
extra : rebase_source : 83ef3414b74a5164b8535bdf266075d1076e0437
2016-11-30 07:22:20 +09:00
Mike Hommey
97ae022e70 Bug 1311039 - Properly detect the default malloc zone on OSX 10.12. r=njn
--HG--
extra : rebase_source : e625875064e051fa931cab5827debe9fd02d91cf
2016-11-03 10:02:06 +09:00
Emanuel Hoogeveen
b324d2ba21 Bug 1315939 - Use FINAL_LIBRARY = 'memory' in jemalloc and mozjemalloc. r=glandium
--HG--
extra : rebase_source : 94d19e08090839537ca5174baa8b37993d108acc
2016-11-10 06:57:00 -05:00
Eric Rahm
2faeed6d55 Bug 1313485 - Convert XPCOM test TestJemalloc to a gtest. r=njn
MozReview-Commit-ID: 5yzn8o33Ne5

--HG--
rename : xpcom/tests/TestJemalloc.cpp => memory/gtest/TestJemalloc.cpp
2016-11-09 10:24:22 -08:00
Ryan VanderMeulen
3445ad743b Bug 1277704 - Update jemalloc 4 to version 4.3.1. r=glandium
--HG--
extra : rebase_source : d8c50b1012085fad8d7bd940abb41d51423bc12c
2016-07-12 10:37:04 -04:00
Emanuel Hoogeveen
e0a81d7503 Bug 1309573 - Part 2: Define MOZ_HAS_MOZGLUE in various places so that the crash reason gets used. r=glandium
--HG--
extra : rebase_source : d55e8dd75469759f333366bdf9662178ff0a2474
2016-11-08 03:53:00 -05:00
Sebastian Hengst
193fc03fc6 Backed out changeset 1b545e55e42a (bug 1313485) for bustage. r=backout
--HG--
rename : memory/gtest/TestJemalloc.cpp => xpcom/tests/TestJemalloc.cpp
2016-11-07 23:46:09 +01:00
Eric Rahm
52ba7f5179 Bug 1313485 - Convert XPCOM test TestJemalloc to a gtest. r=njn
MozReview-Commit-ID: 5yzn8o33Ne5

--HG--
rename : xpcom/tests/TestJemalloc.cpp => memory/gtest/TestJemalloc.cpp
2016-11-07 14:26:25 -08:00
Mike Hommey
97f635df6b Bug 1303232 - Use per-process reduced thread-ids when munging logalloc logs. r=njn
Bug 1300948 added thread-ids to logalloc logs, and logalloc_munge.py was
munging them all as if they were all under the same namespace. But when
filtering munged logs to only contain logs from a given process,
thread-ids then don't necessarily start from 1, which would be the
desired outcome.

So use a different pool of thread-ids for each process.
2016-09-20 13:44:27 +09:00
Mike Hommey
7e7b4f9fe6 Bug 1300974 - Work around race condition leading to deadlock on fork when enabling LogAlloc. r=njn 2016-09-08 07:56:02 +09:00
Mike Hommey
39fe86c4ed Bug 1300948 - Add thread identifier to LogAlloc output. r=njn 2016-09-08 07:55:03 +09:00
Cameron McCormack
072d6a1fbe Bug 1291356 - Enable multiple jemalloc arenas for MOZ_STYLO builds. r=me
--HG--
extra : rebase_source : eaaaf6fd3aad8a99c7032ce55a1d351f2bf80af1
2016-08-26 20:21:48 +09:00
Wes Kocher
50954c6f31 Merge m-c to autoland, a=merge 2016-08-25 17:15:05 -07:00
Emanuel Hoogeveen
8210833803 Bug 1294732 - Back out all of bug 1271165 as it has served its purpose. r=glandium 2016-08-23 08:45:00 -04:00
Andrew McCreight
5772f60634 Bug 1295695 - Add maybe_pod_* methods to InfallibleAllocPolicy. r=glandium
Bug 1207519 added maybe_pod_* methods to the allocation policy
classes, but did not add them to InfallibleAllocPolicy.

I think the idea of these methods is that the callers are explicitly
opting into fallible behavior, so they will deal with any errors that
occur. For instance, in js::HashTable, this is used to try to shrink
the hash table when there are a lot of unused entries. If the shrink
fails, it just continues to use the existing block of memory.

However, having fallible methods in a supposedly infallible class is
weird, so for now, just use the infallible version.

MozReview-Commit-ID: 97D66Z4oLfl

--HG--
extra : rebase_source : 9a3e0b50a5602a8e4772da88c16e1715c25da7df
2016-08-16 11:09:05 -07:00
Ryan VanderMeulen
69113163cf Merge m-c to inbound. a=merge 2016-08-24 09:09:05 -04:00
Andrew McCreight
faa1eb2316 Bug 1295688 - InfallibleAllocPolicy should crash on overflow. r=glandium
Code that uses InfallibleAllocPolicy presumably wants for operations
to always succeed. However, Vector and HashTable can end up detecting
that growing the data structure will fail due to integer overflow, and
then will call reportAllocOverflow() and fail. I think these cases
should crash.

In addition, pod_malloc and pod_realloc should crash rather than
returning NULL when they detect overflow.

This calls mozalloc_abort rather than MOZ_CRASH directly to avoid
circular #includes, because Assertions.h includes nsTraceRefcnt.h
which includes nscore.h which includes mozalloc.h.

MozReview-Commit-ID: 1g99BXLceQI

--HG--
extra : rebase_source : 927d842588c1f85a50a7a1c50a5546d5f688555f
2016-08-16 10:56:14 -07:00
Kan-Ru Chen
b6d880aca1 Bug 1297276 - Rename mfbt/unused.h to mfbt/Unused.h for consistency. r=froydnj
The patch is generated from following command:

  rgrep -l unused.h|xargs sed -i -e s,mozilla/unused.h,mozilla/Unused.h,

MozReview-Commit-ID: AtLcWApZfES


--HG--
rename : mfbt/unused.h => mfbt/Unused.h
2016-08-24 14:47:04 +08:00
Emanuel Hoogeveen
581ecef4dd Bug 1271165 - Part 6: Change the new functions to MFBT_API to export them on OSX. r=jandem
--HG--
extra : rebase_source : cc3d6cd7cec9c822904364ace811d2c0dff98eff
2016-08-16 07:12:00 -04:00
Emanuel Hoogeveen
f7bdbc26ad Bug 1271165 - Part 4: Enable the new functionality in mozjemalloc. r=ehoogeveen 2016-08-12 07:38:00 -04:00
Emanuel Hoogeveen
57b1d1e8f2 Bug 1271165 - Part 2: Hook the new functions up and provide dummy implementations where needed. r=ehoogeveen 2016-08-12 07:37:00 -04:00
Emanuel Hoogeveen
506ca1034e Bug 1271165 - Part 1: Add protected allocation functions and supporting infrastructure to mozjemalloc. r=ehoogeveen 2016-08-12 07:36:00 -04:00
Emanuel Hoogeveen
54e158268e Bug 1271165 - Part 0: Strip trailing blanks from mozjemalloc. r=ehoogeveen 2016-08-12 07:36:00 -04:00
Alexandre Lissy
0af5b943b6 Bug 1284674 - Remove NUWA r=cyu
MozReview-Commit-ID: GyMRNzOBKw6

--HG--
extra : rebase_source : 293af1cd55f2035ce6a99f4ebf144059c32a2b8f
2016-08-02 14:54:00 +02:00
Chris Peterson
6229606616 Bug 1290645 - Remove VS2013 workarounds in memory/. r=glandium 2016-07-28 11:46:30 -07:00
Tom Tromey
5538d692d3 Bug 1286877 - do not set c-basic-offset for python-mode; r=gps
This removes the unnecessary setting of c-basic-offset from all
python-mode files.

This was automatically generated using

    perl -pi -e 's/; *c-basic-offset: *[0-9]+//'

... on the affected files.

The bulk of these files are moz.build files but there a few others as
well.

MozReview-Commit-ID: 2pPf3DEiZqx

--HG--
extra : rebase_source : 0a7dcac80b924174a2c429b093791148ea6ac204
2016-07-14 10:16:42 -06:00
Mike Hommey
63e7e7ede4 Bug 1280578 - Wrap HeapAlloc, HeapFree and HeapReAlloc on Windows. r=njn
The Win32 has non-malloc based heap allocation functions. They are not
widely used in the Gecko code base, but there are a few places that do
use them. They could be replaced with uses of malloc/free/realloc or
new/delete, but there is now an entirely separate problem that
requires wrapping those functions: the Rust static libraries are using
those functions to allocate memory.
2016-07-20 14:04:40 +09:00
Nicholas Nethercote
6f4e7b36f5 Bug 1282185 (part 2) - Fix an assertion failure in DMD. r=erahm.
--HG--
extra : rebase_source : cb2ae2a5a6fdc84bbcaba64759b0dbce1e161249
2016-07-13 19:32:27 +10:00
Nicholas Nethercote
dcf407f2a8 Bug 1282185 (part 1) - Remove a bogus assertion in DMD. r=erahm.
Since bug 1253512 landed, it's possible for DeadBlocks to lack an allocation
stack.

--HG--
extra : rebase_source : 0efc60192ed0992d2f68838d95586cd888765586
2016-07-13 18:10:56 +10:00
Mike Hommey
16529670ac Bug 1284677 - Change how the default OSX malloc zone is found. r=njn
--HG--
extra : rebase_source : b5ea2bfb37047fa9f8d4d833186694407c73adda
2016-07-08 14:48:16 +09:00
Bill McCloskey
308608ab39 Bug 1262671 - Introduce MFBT BufferList class (r=froydnj) 2016-05-27 09:57:40 -07:00
Mike Hommey
8855262939 Bug 1269171 - Change how mozalloc.h is hooked in STL wrappers. r=froydnj
Since the introduction of the STL wrappers, they have included
mozalloc.h, and multiple times, we've hit header reentrancy problems,
and worked around them as best as we could.

Taking a step back, all mozalloc.h does is:
- declare moz_* allocator functions.
- define inline implementations of various operator new/delete variants.

The first only requires the functions to be declared before they are used,
so mozalloc.h only needs to be included before anything that would use
those functions.

The second doesn't actually require a specific order, as long as the
declaration for those functions comes before their use, and they are
either declared in <new> or implicitly by the C++ compiler.

So all in all, it doesn't matter that mozalloc.h is included before the
wrapped STL headers. What matters is that it's included when STL headers
are included. So arrange things such that mozalloc.h is included after
the first wrapped STL header is fully preprocessed (and all its includes
have been included).
2016-05-22 08:32:40 +09:00
Chris Peterson
8a9e2d2bd4 Bug 1272513 - Part 2: Remove redundant -Wshadow CXXFLAGS from moz.build files. r=glandium 2016-05-14 00:54:55 -07:00
Chris Peterson
353ee65255 Bug 1272513 - Part 1: Suppress -Wshadow warnings-as-errors in some directories. r=glandium 2016-05-11 00:00:01 -07:00
Jan Beich
6d9051edd6 Bug 1259537 - Unbreak libc++ build after bug 1245076. r=glandium
MozReview-Commit-ID: 31otXvuDu4A

--HG--
extra : rebase_source : 048a81a540154b672a9249482124d396336acbc4
2016-05-01 09:32:10 +00:00
Ryan VanderMeulen
662ef75681 Bug 1269959 - Update jemalloc 4 to version 4.1.1. r=glandium
--HG--
extra : rebase_source : cf6f490bd2e366cff642d97927c8a54ef4d762a1
2016-05-03 20:26:00 -04:00
Nicholas Nethercote
2511b2c327 Bug 1267550 (part 2) - Rename MOZ_WARN_UNUSED_RESULT as MOZ_MUST_USE. r=froydnj.
It's an annotation that is used a lot, and should be used even more, so a
shorter name is better.

MozReview-Commit-ID: 1VS4Dney4WX

--HG--
extra : rebase_source : b26919c1b0fcb32e5339adeef5be5becae6032cf
2016-04-27 14:16:50 +10:00
Makoto Kato
234456a100 Bug 1091515 - Don't set 64KB page size on aarch64. r=glandium
Linux/aarch64 supports vairous page size configuration.

(From https://www.kernel.org/doc/Documentation/arm64/memory.txt)
- 4KB + 3 levels
- 4KB + 4 levels
- 64KB + 2 leves

So we shouldn't set fixed 64KB page size on aarch64.


MozReview-Commit-ID: 8IdwBMUW4me

--HG--
extra : rebase_source : 2f3d8de3ea118313cb6dc5f48b7532d9b051c1e6
2016-04-05 18:54:42 +09:00
Nicholas Nethercote
b1378bc5e8 Bug 1260329 - Properly escape the frameTable when running |dmd.py --clamp-contents|. r=mccr8.
--HG--
extra : rebase_source : 7a55328abb0a79cfa6de21734ee3d440b7625f79
2016-03-29 13:34:57 +11:00
Eric Rahm
c2edbab476 Bug 1181142 - Part 1: Make the minimum allocation size word sized on all platforms. r=glandium
Bug 691003 made the minimum allocation size word sized for Linux and OS X, we
now need to do a similar change on Windows as well. For Windows the requirement
is 8-bytes on 32-bit and 16-bytes on 64-bit.
2015-07-07 17:36:31 -07:00
Nicholas Nethercote
43969ea737 Bug 1181142 - Part 0: Update DMD tests to handle a minimum alloc size of 16 on Win64. r=erahm. 2016-03-29 16:31:21 +11:00
Nicholas Nethercote
eb9d46b90f Bug 1253512 (part 2) - Aggregate live blocks. r=erahm.
Due to the change in part 1, DMD now prints an entry for every live block,
which increases the output file size significantly in the default case. However,
a lot of those entries are identical and so can be aggregated via the existing
"num" property.

This patch does that, reducing output size by more than half.

--HG--
extra : rebase_source : 6a3709d068f2fb9bbfe3344d7406f05af896380b
2016-03-29 10:41:06 +11:00
Nicholas Nethercote
0f418f3874 Bug 1253512 (part 1) - Overhaul DMD's "sampling". r=erahm.
DMD currently uses a very hacky form of "sampling" by default to avoid
recording stack traces for all blocks. This makes DMD run faster than when it
records all stack traces.

This patch changes the sampling method used; in fact, it avoids "sampling" at
all. The existence of all heap blocks is now recorded exactly, but by default
we only record an allocation stack for each heap block if a Bernoulli trial
succeeds. This choice works well because getting the stack trace is ~100x
slower than recording the block's existence.

Overall, this approach is simpler and it also gives better output -- the choice
of which blocks to record allocation stacks for is mathematically sound, no
stack trace gets blamed for allocations it didn't do, and block counts and
sizes are now always exact.

Other specific things changed.

- All notion of sampling is removed from the various data structures.

- The --sample-below option is removed in favour of --stacks={partial,full}.

- The format of the JSON output file has changed.

- The names of various test files have changed to reflect concept changes.

--HG--
rename : memory/replace/dmd/test/full-empty-cumulative-expected.txt => memory/replace/dmd/test/complete-empty-cumulative-expected.txt
rename : memory/replace/dmd/test/full-empty-dark-matter-expected.txt => memory/replace/dmd/test/complete-empty-dark-matter-expected.txt
rename : memory/replace/dmd/test/full-empty-live-expected.txt => memory/replace/dmd/test/complete-empty-live-expected.txt
rename : memory/replace/dmd/test/full-unsampled1-dark-matter-expected.txt => memory/replace/dmd/test/complete-full1-dark-matter-expected.txt
rename : memory/replace/dmd/test/full-unsampled1-live-expected.txt => memory/replace/dmd/test/complete-full1-live-expected.txt
rename : memory/replace/dmd/test/full-unsampled2-cumulative-expected.txt => memory/replace/dmd/test/complete-full2-cumulative-expected.txt
rename : memory/replace/dmd/test/full-unsampled2-dark-matter-expected.txt => memory/replace/dmd/test/complete-full2-dark-matter-expected.txt
rename : memory/replace/dmd/test/full-sampled-live-expected.txt => memory/replace/dmd/test/complete-partial-live-expected.txt
extra : rebase_source : 47d287405dc5e9075f08addaba49e879c2c6e23f
2016-02-24 14:42:22 +11:00
Sebastian Hengst
554c274197 Backed out changeset 4593c32d5b81 (bug 1181142) for making XPCshell test_dmd.js fail on Windows 8 x64 debug. r=backout 2016-03-25 21:53:30 +01:00
Eric Rahm
9386c16326 Bug 1181142 - Part 1: Make the minimum allocation size word sized on all platforms. r=glandium
Bug 691003 made the minimum allocation size word sized for Linux and OS X, we
now need to do a similar change on Windows as well. For Windows the requirement
is 8-bytes on 32-bit and 16-bytes on 64-bit.
2015-07-07 17:36:31 -07:00
Nicholas Nethercote
5752d3b3e6 Bug 1258257 - Reduce mozjemalloc page cache size from 4 MiB to 1 MiB. r=glandium.
This reduces memory usage by up to 3 MiB per process.

MozReview-Commit-ID: Gfs9PIJM4br

--HG--
extra : rebase_source : 69ac5acf7f7f0c802a047f5bf72838e5c4d1f123
2016-03-24 16:38:45 +11:00
Ted Mielczarek
68bd5e8c4d bug 1244743 - Replace MOZ_NATIVE_X with MOZ_SYSTEM_X. r=gps
MozReview-Commit-ID: 9ip3qeAXFEe

--HG--
extra : commitid : H6aEkHprVyX
extra : rebase_source : e20d5f4b297caf66711c72cd55dd76f7f9ef9d9c
extra : histedit_source : 47545e0ea61eb66f16545f70e6d3792a389bedb1
2016-02-01 10:49:34 -05:00
Mike Hommey
8170c2d9bc Bug 1254850 - Update jemalloc 4 to version 4.1.0. r=njn 2016-03-11 09:38:28 +09:00
Ms2ger
d1f112a81b Bug 1254948 - Remove makefiles used to transition from jemalloc_config.c to jemalloc_config.cpp; r=glandium 2016-03-10 15:53:58 +01:00
Jonathan Watt
a9c7d25928 Bug 1253094, part 10 - Stop using DebugOnly for class/struct members in memory/. r=njn
MozReview-Commit-ID: LLSCgDES38Q
2016-02-26 15:52:08 +00:00