Commit Graph

537 Commits

Author SHA1 Message Date
Mike Hommey
03b3291d3f Bug 1120272 - Expose _aligned_alloc and _aligned_free from mozglue. r=njn 2015-01-20 13:10:00 +09:00
Seth Fowler
fb673b7c3a Bug 1121297 (Part 2) - Make VolatileBuffer threadsafe. r=glandium 2015-01-16 15:47:35 -08:00
Seth Fowler
26bd4c1cb6 Bug 1121297 (Part 1) - Move VolatileBuffer into libxul. r=glandium
--HG--
rename : memory/mozalloc/VolatileBuffer.h => memory/volatile/VolatileBuffer.h
rename : memory/mozalloc/VolatileBufferAshmem.cpp => memory/volatile/VolatileBufferAshmem.cpp
rename : memory/mozalloc/VolatileBufferFallback.cpp => memory/volatile/VolatileBufferFallback.cpp
rename : memory/mozalloc/VolatileBufferOSX.cpp => memory/volatile/VolatileBufferOSX.cpp
rename : memory/mozalloc/VolatileBufferWindows.cpp => memory/volatile/VolatileBufferWindows.cpp
rename : memory/mozalloc/moz.build => memory/volatile/moz.build
rename : memory/mozalloc/tests/TestVolatileBuffer.cpp => memory/volatile/tests/TestVolatileBuffer.cpp
rename : memory/mozalloc/tests/moz.build => memory/volatile/tests/moz.build
2015-01-16 15:47:22 -08:00
Nicholas Nethercote
a479e9ccad Bug 1121830 - DMD: add "num" property to blocks in the output. r=mccr8.
The new "num" property lets identical blocks be aggregated in the output. This
patch only uses the "num" property for dead blocks, because that's where the
greatest potential benefit lies, but it could be used for live blocks as well.

On one test case (a complex PDF file) running with --mode=cumulative
--sample-below=1 this patch had the following effects.

- Change in running speed was negligible.

- Compressed output file size dropped from 8.8 to 5.0 MB.

- Compressed output file size dropped from 297 to 50 MB.

- dmd.py runtime (without stack fixing) dropped from 30 to 8 seconds.

--HG--
extra : rebase_source : 46a32058cd5c31cd823fe3f1accb5e68bcd320f3
2015-01-15 20:38:38 -08:00
Nicholas Nethercote
059cf96c41 Bug 1122337 - Mark DMD directories as FAIL_ON_WARNINGS. r=glandium.
The DEFINES and XPCOM_API changes are needed to get rid of "inconsistent dll
linkage" warnings on Windows builds.

--HG--
extra : rebase_source : 00756f51ebee85c70f65d51dbac17b4835262697
2015-01-15 21:17:18 -08:00
Mike Hommey
af38583e2e Bug 1121314 - Avoid needing the arena in chunk_alloc_default to avoid possible infinite loops involving a0malloc. r=njn
The infinite loop happens if chunk_alloc_arena needs to be called when a0malloc
is called. It in turn calls chunk_alloc_default, which uses tsd, which calls
a0malloc if it's the first time the tsd is being gotten from the current thread.
tsd only uses a0malloc on platforms where there is no native thread local storage
support, which, for Mozilla, essentially means anything that is not Linux.

But the tsd is only neededto get the dss precedence setting of the given arena.
That setting has no effect when dss is disabled, which it is on Windows and Mac.

Moreover, the default setting for dss precedence is "secondary", which means
jemalloc only tries dss after it failed to get memory with mmap/VirtualAlloc.
Considering the cases where mmap/VirtualAlloc would fail essentially means
there is shortage of address space, sbrk() is not going to have much more
success, so we might as well disable dss support on all platforms, avoiding
the infinite loop problem on Android and B2G as well.
2015-01-14 15:18:49 +09:00
Mike Hommey
2e083509f2 Bug 1120798 - Use a larger realloc in one of the DMD tests so that it passes for mozjemalloc as well as jemalloc3. r=njn 2015-01-14 15:18:49 +09:00
Masatoshi Kimura
bf312ad056 Bug 1120062 - Part 1: Remove most Nullptr.h includes. r=waldo 2015-01-11 11:34:52 +09:00
Brad Lassey
0e7333b6e7 bug 1083116 - build fails: malloc_decls.h conflicting types for 'malloc_usable_size' r=glandium 2015-01-06 19:58:36 -05:00
Nicholas Nethercote
cdd9ba9d57 Bug 1088343 (part 2) - Enable test_dmd on Win64. r=glandium.
--HG--
extra : rebase_source : b85a5e35c71f44d7f615bc67863711a4876e9da4
2015-01-08 16:04:43 -08:00
Ehsan Akhgari
4354953b4f Bug 1118486 - Part 1: Use = delete instead of MOZ_DELETE directly; r=Waldo
Most of this patch (with the exception of dom/bindings/Codegen.py) was
generated by the following bash script:

#!/bin/bash

function convert() {
echo "Converting $1 to $2..."
find . ! -wholename "*nsprpub*" \
       ! -wholename "*security/nss*" \
       ! -wholename "*/.hg*" \
       ! -wholename "*/.git*" \
       ! -wholename "obj-*" \
         -type f \
      \( -iname "*.cpp" \
         -o -iname "*.h" \
         -o -iname "*.cc" \
         -o -iname "*.idl" \
         -o -iname "*.ipdl" \
         -o -iname "*.ipdlh" \
         -o -iname "*.mm" \) | \
    xargs -n 1 sed -i -e "s/\b$1\b/$2/g"
}

convert MOZ_DELETE '= delete'
2015-01-08 23:19:05 -05:00
Guilherme Goncalves
c6a10e7bb5 Bug 899126 - Part 2 - Measure bookkeeping in jemalloc3. r=glandium. 2014-12-29 13:42:14 -08:00
Guilherme Goncalves
750b113679 Bug 899126 - Part 1.5 - Implement bookkeeping for jemalloc3. r=glandium 2014-12-29 13:37:24 -08:00
Jan Beich
f4828d35f5 Bug 1115078 - Unbreak MOZ_JEMALLOC3 build on FreeBSD after bug 899126. r=glandium 2014-12-23 12:32:00 -05:00
Guilherme Goncalves
1fb7597a1b Bug 1110514 - Use arena.<i>.purge to free jemalloc3's dirty pages. r=glandium 2014-12-23 02:41:00 -05:00
Guilherme Goncalves
802cb55877 Bug 1108045 - Junk memory with jemalloc3. r=glandium 2014-12-23 07:44:00 -05:00
Guilherme Goncalves
f9d70a5255 Bug 762448 - Bump opt.lg_dirty_mult in jemalloc3 to reduce number of dirty pages. r=glandium 2014-12-23 06:01:00 -05:00
Guilherme Goncalves
5ff0106324 Bug 762448 - Deduct bin_unused from waste in jemalloc3's stats. r=glandium 2014-11-12 13:00:00 -05:00
Ryan VanderMeulen
42a3e6c0ea Merge m-c to inbound. a=merge 2014-12-23 13:18:02 -05:00
Guilherme Goncalves
c41a18c8d2 Bug 1112566 - Add a isblank definition for MSVC < 2013. r=glandium 2014-12-23 15:27:19 +09:00
Guilherme Goncalves
aa02623d99 Bug 1112566 - Move variable declaration to the top its block for MSVC compatibility. r=jasone,r=glandium 2014-12-23 15:27:18 +09:00
Vladimir Vukicevic
a910628209 Bug 1112566 - Use jemalloc C99 msvc compat headers when building mozjemalloc compat glue and jemalloc3 replace-malloc library. r=glandium 2014-12-23 15:27:18 +09:00
Guilherme Goncalves
22c6a4800b Bug 899126 - Part 1: Implement bin_unused for jemalloc3. r=glandium 2014-12-19 09:15:00 -05:00
Nicholas Nethercote
53da6fb716 Bug 1112968 - dmd.py: Tweak the --alloc-fns list for better effect on Windows. r=mccr8.
--HG--
extra : rebase_source : 3dc8a09f9a22bf234406c0ff89d7b4050c7908e7
2014-12-18 18:59:53 -08:00
Nicholas Nethercote
8030658f62 Bug 1112968 - dmd.py: Tweak the --alloc-fns list for better effect on Mac. r=mccr8.
--HG--
extra : rebase_source : 03dbab1eeb4e3f99672a174277ba634a39a98292
2014-12-17 19:10:59 -08:00
Guilherme Goncalves
60eb32a82c Bug 1094275 - Update the in-tree copy of jemalloc3 to commit b4acf73. r=glandium 2014-12-17 09:14:48 +09:00
Nicholas Nethercote
658df429fc Bug 1110455 - Add --sort-by=num-blocks option to dmd.py. r=mccr8.
--HG--
extra : rebase_source : 6d6aa87f102ee069f4f5052607463380c656f80d
2014-12-15 15:04:39 -08:00
Nicholas Nethercote
e358318e29 Bug 1111419 - dmd.py: add a cache for recordKeyParts. r=mccr8.
This reduces the runtime on my Linux machine for one large DMD output file from
235 seconds to 105 seconds.

--HG--
extra : rebase_source : bd567319c1207a09fa47b94571066109de1cc9e7
2014-12-15 15:04:34 -08:00
Ehsan Akhgari
ebdf8e7c36 Bug 1109746 - Fix more bad implicit constructors in some random code across the tree
--HG--
extra : amend_source : 48ca151cb1ca2193404c65831556051c02093914
2014-12-13 14:17:04 -05:00
Ryan VanderMeulen
9a56a97e41 Backed out changesets 34a66a625618, 32db3daa20fe, and ac15ba8568c5 (bug 1102219) for Gaia unit test failures.
--HG--
extra : rebase_source : 0266cc0b2a6568ca653885745f381df8d6d167ec
2014-12-11 11:28:30 -05:00
ziyunfei
5f470bd4a1 Bug 1102219 - Part 3: Replace String.prototype.contains with String.prototype.includes in chrome code. r=till 2014-12-10 20:50:00 -05:00
Chris Peterson
793516eb34 Bug 1108934 - Fix -Wunused-const-variable warning-as-error in non-unified build of memory/mozalloc. r=glandium 2014-12-08 21:26:27 -08:00
Nicholas Nethercote
ee6de8d6c9 Bug 1100851 - Print the mode in dmd.py's output. r=mccr8.
--HG--
extra : rebase_source : ae0f985ba8a74dfec8a36ff1a380279ed18b675b
2014-12-10 13:33:50 -08:00
Nicholas Nethercote
98688c1a41 Bug 1100851 - Tweak DMD to account for the fact that $DMD can now be undefined. r=glandium.
Now that defining $DMD is no longer necessary to run DMD, this patch does the
following.
- Removes all the places where we set DMD=1 (test harnesses, etc.)
- Still handles DMD=1, for backwards compatibility.
- Prints "$DMD is undefined" at DMD start-up if appropriate.
- Writes a |null| value for |dmdEnvVar| in the JSON if $DMD is undefined. Bumps
  the DMD output version number accordingly.
- Changes a bunch of the test files accordingly, including changing the mode of
  script-ignore-alloc-fns.json in order to test a case where $DMD is undefined.

--HG--
extra : rebase_source : eb1ef5722410734ce6d7658465ff6f442ee4ed49
2014-11-27 21:04:46 -08:00
Mike Hommey
548d773d4e Bug 1104634 - Force optimize mozjemalloc on --disable-optimize builds. r=gps
The reason for --disable-optimize is to make debugging easier, but not many
people actually need a high level of debuggability of the allocator itself.
This works around the issue that the Android NDK's definition of ffs is
broken when compiling without optimization, while avoiding to add yet another
configure test.
2014-12-10 10:50:44 +09:00
Nicholas Nethercote
5e166635cb Bug 1094552 (part 6) - DMD: add support for cumulative heap profiling. r=mccr8.
By adding a new "cumulative" mode.

--HG--
extra : rebase_source : 5c851b7c594a134fae48393ff0becfd057715041
2014-10-30 20:22:47 -07:00
Nicholas Nethercote
c3648686f0 Bug 1094552 (part 5) - DMD: choose the profiling mode at start-up. r=mccr8.
This patch moves profiling mode selection from post-processing (in dmd.py) to
DMD start-up. This will make it easier to add new kinds of profiling, such as
cumulative heap profiling.

Specifically:

- There's a new --mode option. |LiveWithReports| is the default, as it is
  currently.

- dmd.py's --ignore-reports option is gone.

- There's a new |mode| field in the JSON output.

- Reports-related operations are now no-ops if DMD isn't in LiveWithReports
  mode.

- Diffs are only allowed for output files that have the same mode.

- A new function ResetEverything() replaces the SetSampleBelowSize() and
  ClearBlocks(), which were used by the test to change DMD options.

- The tests in SmokeDMD.cpp are split up so they can be run multiple times, in
  different modes. The exact combinations of tests and modes has been changed a
  bit.

--HG--
rename : memory/replace/dmd/test/full-reports-empty-expected.txt => memory/replace/dmd/test/full-empty-dark-matter-expected.txt
rename : memory/replace/dmd/test/full-heap-empty-expected.txt => memory/replace/dmd/test/full-empty-live-expected.txt
rename : memory/replace/dmd/test/full-heap-sampled-expected.txt => memory/replace/dmd/test/full-sampled-live-expected.txt
rename : memory/replace/dmd/test/full-reports-unsampled1-expected.txt => memory/replace/dmd/test/full-unsampled1-dark-matter-expected.txt
rename : memory/replace/dmd/test/full-heap-unsampled1-expected.txt => memory/replace/dmd/test/full-unsampled1-live-expected.txt
rename : memory/replace/dmd/test/full-reports-unsampled2-expected.txt => memory/replace/dmd/test/full-unsampled2-dark-matter-expected.txt
rename : memory/replace/dmd/test/script-diff-basic-expected.txt => memory/replace/dmd/test/script-diff-live-expected.txt
rename : memory/replace/dmd/test/script-diff1.json => memory/replace/dmd/test/script-diff-live1.json
rename : memory/replace/dmd/test/script-diff2.json => memory/replace/dmd/test/script-diff-live2.json
extra : rebase_source : bf32cc4e0d82aa1a20ceb55e8ea259850b49cc06
2014-12-08 14:45:14 -08:00
Nicholas Nethercote
7a1087c6c6 Bug 1094552 (part 4) - Rename AnalyzeReports() as Analyze(). r=mccr8.
Because DMD is no longer just about measuring memory reports coverage, but is
also used for more general heap profiling.

--HG--
extra : rebase_source : 82b4579de240037f96cf6618b15870925adc431b
2014-12-08 14:45:13 -08:00
Nicholas Nethercote
e4c12e978c Bug 1102525 (part 1) - Add InfallibleAllocPolicy to mozalloc. r=glandium.
--HG--
extra : rebase_source : a79162fb7f73e52c5c8df29c8229efc64c3451e5
2014-12-08 14:45:10 -08:00
Chris Peterson
914cf9e473 Back out part of changeset 2147895c5dc4 (bug 1107814) because of non-unified OS X build breakage on a CLOSED TREE. a=philor 2014-12-08 21:42:58 -08:00
Chris Peterson
e424225899 Bug 1107814 - Part 2: Mark some directories as FAIL_ON_WARNINGS conditionally for some compilers. r=gps 2014-12-04 01:11:14 -08:00
Mike Hommey
c04636bda3 Bug 1104173 - Properly export jemalloc_* symbols from the jemalloc3 replace-malloc library. r=njn
Bug 818922 made MOZ_REPLACE_MALLOC a global define, and that changed the
defines set when building mozjemalloc_compat.c for the jemalloc3
replace-malloc library. In turn, this made the symbol munging wrong for
that library, making the jemalloc_* functions exported as je_jemalloc_*
instead of replace_jemalloc_*.
2014-11-26 16:04:25 +09:00
Nicholas Nethercote
5fad01bb19 Bug 1094552 (part 3) - DMD: rename |Block| as |LiveBlock|. r=mccr8.
This is to give better contrast with |DeadBlock|, which will be added in the
next patch.

--HG--
extra : rebase_source : cbc767fcc5667cfed108ca7c4ebf1d7e82aa185e
2014-11-18 18:57:20 -08:00
Nicholas Nethercote
29bf26b228 Bug 1094552 (part 2) - DMD: tweak Range and Enum use. r=mccr8.
This patch:

- Uses |auto| in Range loops, so more of them fit on a single line.

- Converts one use of HashSet::Enum (which is only needed if you're modifying
  the HashSet as you iterate) to HashSet::Range.

--HG--
extra : rebase_source : 09b011cd69218c06984f06420d375839cd4e9214
2014-11-23 17:39:17 -08:00
Nicholas Nethercote
f38abd1199 Bug 1094552 (part 1) - DMD: rename mSampled field. r=mccr8.
The getter is called IsSampled(), so the field should match that.

--HG--
extra : rebase_source : 074c748eff909faee51bc03888b0c7f328bff82e
2014-11-18 19:11:36 -08:00
Emanuel Hoogeveen
d573431119 Bug 1100485 - Lock chunks during double purging to avoid racing with allocation. r=glandium 2014-11-19 09:34:00 +01:00
Mike Hommey
e68882a91d Bug 1098967 - Use native Win32 APIs to append to the LogAlloc output. r=nfroyd
Because it turns out that the POSIX API the CRT exposes doesn't do O_APPEND
in a sane manner.
2014-11-19 09:46:49 +09:00
Mike Hommey
558fe59234 Bug 1087245 part 2 - Register the MALLOC_LOG file handle from LogAlloc to poison IO interposer. r=nfroyd 2014-11-18 19:21:06 +09:00
Mike Hommey
3cbc3cf69f Bug 1087245 part 1 - Allow replace-malloc libraries to register debug file handles to poison IO interposer. r=nfroyd 2014-11-18 19:21:06 +09:00
Mike Hommey
7def0f9cd3 Bug 1097507 - Make libxul independent of libdmd when DMD is enabled. r=njn
This also effectively changes how DMD is enabled from requiring both
replace-malloc initialization and the DMD environment variable to
requiring only the former. The DMD environment variable can still be
used to specify options, but not to disable entirely.

This however doesn't touch all the parts that do enable DMD by setting
the DMD environment variable to 1, so the code to handle this value
is kept.
2014-11-18 19:21:06 +09:00