Tag:
Branch:
Tree:
fd9fbe7451
lts_2018_06_20
lts_2018_12_18
lts_2019_08_08
lts_2020_02_25
lts_2020_09_23
lts_2021_03_24
lts_2021_11_02
lts_2022_06_23
master
20180600
20181200
20181200.1
20190808
20190808.1
20200225
20200225.1
20200225.2
20200225.3
20200923
20200923.1
20200923.2
20200923.3
20210324.0
20210324.1
20210324.2
20210324.rc1
20211102.0
20211102.rc2
20220623.0
20220623.1
20220623.rc1
${ noResults }
27 Commits (fd9fbe74511fb2a060a3b15a1572b52fe16a6a43)
| Author | SHA1 | Message | Date |
|---|---|---|---|
Gennadiy Rozental
|
e225d0b2c6 |
No changes in OSS
PiperOrigin-RevId: 478668020 Change-Id: Iee79011bf7154e83ad862e7bf6e7a76dd337ec06 |
2 years ago |
Martijn Vels
|
b707b6c179 |
Add GetCustomAppendBuffer method to absl::Cord
The Cord::GetCustomAppendBuffer() method provides the same functionality as Cord::GetAppendBuffer(), except that callers can specify a custom block size/limit if the method returns a newly allocated buffer. In other words: Cord::GetAppendBuffer() defaults to CordBuffer::CreateWithDefaultLimit(), Cord::GetCustomAppendBuffer() defaults to CordBuffer::CreateWithCustomLimit(). PiperOrigin-RevId: 461231989 Change-Id: I5c03f31139d9b068feee1bea76d59e1c5e30ef07 |
2 years ago |
Derek Mauro
|
48419595d3 |
Release absl::CordBuffer
absl::CordBuffer holds data for eventual inclusion within an existing absl::Cord. CordBuffers are useful for building large Cords that may require custom allocation of its associated memory, a pattern that is common in zero-copy APIs. PiperOrigin-RevId: 453212229 Change-Id: I6a8adc3a8d206691cb1b0001a9161e5080dd1c5f |
3 years ago |
Benjamin Barenblat
|
3af16ffea8 |
Fix an msan warning in cord_ringbuffer_test
Stop the absl::Cord destructor from running on the constinit cord in CordTest.ConstinitConstructor. This allows inspecting the cord at any point while the test is exiting, which is important for the semantics of the test. PiperOrigin-RevId: 448327386 Change-Id: Icef9faa2b63f1f0ae60b3430dcf6184f5dead885 |
3 years ago |
Abseil Team
|
0ad7994f10 |
Export of internal Abseil changes
-- 91d76b3ac9edff91f206d9eee60423c39eeeaf93 by Derek Mauro <dmauro@google.com>: Internal change PiperOrigin-RevId: 430442277 -- 9f8a87bcc5cc5b0fd8b7f0318f37d152fd8bea06 by Evan Brown <ezb@google.com>: Small refactoring to work towards allowing for node_btree_* containers. - Change common_params::transfer to rely on slot_policy transfer instead of manually doing construct/destruct. Transfer is not construct/destruct for node containers. - Move maps' value_compare into btree.h from btree_map.h so it can be reused for node_btree_map.h. - Also add a test for maps' value_compare protected members. PiperOrigin-RevId: 430245542 -- 0126e0b6295342317d9c9f0a66e2d7009b858426 by Martijn Vels <mvels@google.com>: Add CordRepSubString::Create function with hard checks on IsFlat() | IsExternal() This hardens internal invariants, IsFlat() || IsExternal() is a cheap, single predicted branch, and removes boilerplate code from cord.cc PiperOrigin-RevId: 429676041 -- ed98a92af49d9e238d9f1d1b69fb4eddcd1ccbc7 by Abseil Team <absl-team@google.com>: tweaks to status.h documentation to reflect general-purpose communication PiperOrigin-RevId: 429584104 GitOrigin-RevId: 91d76b3ac9edff91f206d9eee60423c39eeeaf93 Change-Id: I54d6d116a564f86a842b983ca76559bf9b388f72 |
3 years ago |
|
Abseil Team
|
7f850b3167 |
Export of internal Abseil changes
-- ed829ac612f090375427c3488827c6e74deb2e3f by Derek Mauro <dmauro@google.com>: Update latest GCC/Clang Linux tests to Bazel 5.0.0 and CMake 3.22.2 PiperOrigin-RevId: 429369775 -- 76952303c4d942288c4e7657ffb5893cec54a132 by Martijn Vels <mvels@google.com>: Optimize Cord::ChunkIterator now that CordRepConcat is removed PiperOrigin-RevId: 429321455 -- dcd0d287793649aba9b98268c5783e449a34749f by Martijn Vels <mvels@google.com>: Add IsDataEdge() and DataEdgeValue() helper functions. This moves repetitive logic accessing data edges into its own header, and more strongly defines the notion of what a data edge is, enforcing the internal invariants. This will also be incorporated in optimized Cord iteration logic once CordRepConcat is totally removed from the Cord code. PiperOrigin-RevId: 429307248 -- 6a0903962155988085bf8656743fda9c4cdcba6c by Abseil Team <absl-team@google.com>: Make it clear that the probability function given for the zipf distribution is unnormalized, i.e., sum(p(x) for x = 0..k) != 100%. Quoting Section 7 of the paper cited in the comments, where this formula comes from (emphasis mine): "We will consider the two parameter generalization as defined in Dagpunar [1988] with the *unnormalized* probability function ..." PiperOrigin-RevId: 429068258 -- 3899ff6d444ba755148bc521a6ee031d9e9d4485 by Abseil Team <absl-team@google.com>: Internal Changes PiperOrigin-RevId: 428644856 -- 319de702d2b537cbb76c4c71277ae89b349b162e by Benjamin Barenblat <bbaren@google.com>: Support symbolization on PA-RISC Null out supervisor bits in PA-RISC addresses before symbolizing, and handle function descriptor tables correctly. Change symbolize_test.cc to use 32-bit aligned addresses, allowing that test to pass on PA-RISC. PiperOrigin-RevId: 428590564 GitOrigin-RevId: ed829ac612f090375427c3488827c6e74deb2e3f Change-Id: Ie01ff3b9365fd45e5a55f858038552679f3180d3 |
3 years ago |
|
Abseil Team
|
36db0e4b69 |
Export of internal Abseil changes
--
4409b08e103d6e7041d18a4d431290cafe3650cf by Derek Mauro <dmauro@google.com>:
Workaround NVCC compile error in StringConstant
Based on a patch in TensorFlow:
|
3 years ago |
|
Abseil Team
|
52d41a9ec2 |
Export of internal Abseil changes
-- 81f95fcf85b75b84f9892c73123501472b9cff33 by Martijn Vels <mvels@google.com>: Introduce GetEstimatedMemoryUsage(CordMemoryAccounting::kFairShare) Memory usage analysis is moved into a separate cord_analysis.cc source. PiperOrigin-RevId: 416370158 -- 6bc7b1348fd27fe53f100c9eabd47f4f2cb9c19c by Abseil Team <absl-team@google.com>: Support scoped enum in absl::Substitute. PiperOrigin-RevId: 416345422 -- 6399f4f6ae05ebcd67664ebd844902f699ab8ec7 by Abseil Team <absl-team@google.com>: Correct the computation of contention cycles Currently, we record contention cycles from the first time a thread started waiting on a mutex. Consider a situation in which two threads, T1 and T2, run a loop at the top of which they acquire a common mutex and release it at the end of the loop body. Further assume that T2 is never able to acquire the mutex as T1 repeatedly acquires and then releases the mutex. In this case, we would expect that the reported contention cycles would be increase linearly over time. But currently we observe a quadratic behavior in the reported waiting time as mentioned in b/14684244#comment10. To fix the issue, this CL records the contention cycles experienced by all the threads woken up when the mutex is released. Further, contention_start_cycles is set to the current time since the contention cycles for the time already passed has been taken into account. With this CL, we get a linear increase in the waiting time, the expected behavior. PiperOrigin-RevId: 416322593 -- 149c1637c8a0f1a38e5a8f9f27e5803a2015a554 by Jorg Brown <jorg@google.com>: Make Status::EmptyString more efficient by constructing it in global space, rather than on the heap. See https://godbolt.org/z/8M9n7YqcY for reduced code size. PiperOrigin-RevId: 416307833 -- 3b4562a8be5a3c80077cb67b0a32c97419058380 by Abseil Team <absl-team@google.com>: Clarify the usage of RegisterMutexProfiler PiperOrigin-RevId: 416146130 GitOrigin-RevId: 81f95fcf85b75b84f9892c73123501472b9cff33 Change-Id: Iccb72d7ee617e6ebe226a38170d62e0849b43480 |
3 years ago |
|
Abseil Team
|
1065514ef3 |
Export of internal Abseil changes
--
07240ca7822d007cdcc79f2c40bd58b2c2010348 by Abseil Team <absl-team@google.com>:
Correct the comment from "AlphaNum" to "Arg".
PiperOrigin-RevId: 416139192
--
adcba4a6b3763626e1db7b1e8c108b3114903557 by Martijn Vels <mvels@google.com>:
Fix NewExternalRep() to require data being non-empty, and remove nullptr return.
PiperOrigin-RevId: 416135865
--
c0d14cd918fb16f15d1d84de9284b5c5ecc1f8f2 by Abseil Team <absl-team@google.com>:
Fix doc comment for absl::ascii_isprint().
The comment was incorrectly saying that it includes all whitespace.
It doesn't; the only whitespace char it includes is ' '.
PiperOrigin-RevId: 416112524
--
d83327800159c07002b6865e21232a12463e02dd by Abseil Team <absl-team@google.com>:
Internal change
PiperOrigin-RevId: 416099978
--
baf11e9ca42ca9140cdbf8075f971db8d65b1195 by Ilya Tokar <tokarip@google.com>:
Prevent compiler from optimizing Group_Match* benchmarks away.
Currently we benchmark single store of precomputed value.
Not all affected benchmarks show performance changes:
BM_Group_Match 0.53ns ± 1% 0.53ns ± 0% -0.42% (p=0.038 n=10+10)
BM_Group_MatchEmpty 0.26ns ± 1% 0.26ns ± 1% ~ (p=1.000 n=10+10)
BM_Group_MatchEmptyOrDeleted 0.26ns ± 1% 0.26ns ± 1% ~ (p=0.121 n=10+10)
BM_Group_CountLeadingEmptyOrDeleted 0.26ns ± 1% 0.45ns ± 0% +70.05% (p=0.000 n=10+8)
BM_Group_MatchFirstEmptyOrDeleted 0.26ns ± 0% 0.44ns ± 1% +65.91% (p=0.000 n=8+9)
But inspecting the generated code shows the difference,
e. g. BM_Group_MatchFirstEmptyOrDeleted
Before:
add $0xffffffffffffffff,%rbx
jne 30
After:
pcmpeqd %xmm0,%xmm0
pcmpgtb -0x30(%rbp),%xmm0
pmovmskb %xmm0,%eax
add: 0x23$0xffffffffffffffff,%rbx
jne 40
PiperOrigin-RevId: 416083515
--
122fbff893dc4571b3e75e4b241eb4495b925610 by Abseil Team <absl-team@google.com>:
Put namespace guard in ABSL_DECLARE_FLAG to make declaring a flag in a namespace a compiler error instead of a linker error.
PiperOrigin-RevId: 416036072
--
020fd8a20f5fa319e948846e003391fcb9e03868 by Ilya Tokar <tokarip@google.com>:
Make Cord::InlineRep::set_data unconditionally zero out memory.
Currently there is a single case where we don't zero out memory
as an optimization. Unconditional zeroing doesn't show any changes
in benchmarks, except for the unrelated improvement:
BM_CordPartialCopyToCord/1M/1 12.6ns ± 4% 12.6ns ± 4% ~ (p=0.857 n=16+19)
BM_CordPartialCopyToCord/1M/128 44.9ns ± 7% 45.0ns ± 3% ~ (p=0.468 n=18+17)
BM_CordPartialCopyToCord/1M/1k 64.5ns ± 4% 61.4ns ± 4% -4.82% (p=0.000 n=19+17)
BM_CordPartialCopyToCord/1M/8k 139ns ± 3% 128ns ±15% -7.76% (p=0.009 n=17+20)
BM_CordPartialCopyToCord/1M/16k 193ns ± 6% 168ns ± 6% -13.17% (p=0.000 n=17+17)
BM_CordPartialCopyToCord/4M/16k 199ns ± 4% 177ns ± 4% -11.36% (p=0.000 n=17+18)
BM_CordPartialCopyToCord/4M/32k 275ns ± 3% 250ns ± 4% -9.00% (p=0.000 n=18+18)
BM_CordPartialCopyToCord/4M/64k 291ns ± 4% 266ns ± 5% -8.53% (p=0.000 n=18+16)
BM_CordPartialCopyToCord/4M/128k 322ns ± 5% 291ns ± 4% -9.43% (p=0.000 n=20+18)
BM_CordPartialCopyToCord/8M/32k 281ns ± 5% 251ns ± 4% -10.38% (p=0.000 n=20+16)
BM_CordPartialCopyToCord/8M/64k 293ns ± 6% 267ns ± 4% -8.87% (p=0.000 n=16+19)
BM_CordPartialCopyToCord/8M/128k 334ns ± 3% 305ns ± 2% -8.56% (p=0.000 n=17+16)
This is clearly an alignmnet effect since number of the executed instructions is the same:
M_CordPartialCopyToCord/1M/1 155 ± 0% 155 ± 0% ~ (all samples are equal)
BM_CordPartialCopyToCord/1M/128 446 ± 0% 446 ± 0% ~ (p=0.332 n=36+39)
BM_CordPartialCopyToCord/1M/1k 473 ± 0% 473 ± 0% ~ (p=0.969 n=40+40)
BM_CordPartialCopyToCord/1M/8k 808 ± 0% 808 ± 0% ~ (p=0.127 n=40+39)
BM_CordPartialCopyToCord/1M/16k 957 ± 0% 957 ± 0% ~ (p=0.532 n=40+40)
BM_CordPartialCopyToCord/4M/16k 952 ± 0% 952 ± 0% ~ (p=0.686 n=39+39)
BM_CordPartialCopyToCord/4M/32k 1.12k ± 0% 1.12k ± 0% ~ (p=0.690 n=40+40)
BM_CordPartialCopyToCord/4M/64k 1.23k ± 0% 1.23k ± 0% ~ (p=0.182 n=40+39)
BM_CordPartialCopyToCord/4M/128k 1.44k ± 0% 1.44k ± 0% ~ (p=0.711 n=40+40)
BM_CordPartialCopyToCord/8M/32k 1.12k ± 0% 1.12k ± 0% ~ (p=0.697 n=40+40)
BM_CordPartialCopyToCord/8M/64k 1.23k ± 0% 1.23k ± 0% +0.00% (p=0.049 n=40+40)
BM_CordPartialCopyToCord/8M/128k 1.44k ± 0% 1.44k ± 0% ~ (p=0.507 n=40+40)
This makes code simpler and doesn't regress performance.
PiperOrigin-RevId: 415560574
--
37305b2690b31682088749e4d62f40d7095bdc54 by Derek Mauro <dmauro@google.com>:
Internal change
PiperOrigin-RevId: 415558737
--
86aaed569b9e743c1eb813a5f48def978a793db3 by Martijn Vels <mvels@google.com>:
Internal change
PiperOrigin-RevId: 415515201
--
6cdb8786cdcb4fa0b8a4b72fc98940877d1fdeff by Abseil Team <absl-team@google.com>:
Update SubmitMutexProfileData to accept wait_cycles instead of wait_timestamp
PiperOrigin-RevId: 415360871
--
9f979d307aa16ad09f214e04876cbe84395c0901 by Abseil Team <absl-team@google.com>:
absl::flat_hash_set compiles with -Wconversion -Wsign-compare
PiperOrigin-RevId: 415357498
--
9eceb14174708f15e61259d449b214a8a4c7f9e7 by Abseil Team <absl-team@google.com>:
Fix AddressIsReadable for the corner case of (aligned) addr == NULL.
PiperOrigin-RevId: 415307792
--
1a39ffe55898375e2d7f88c17c99db5a1b95b313 by Martijn Vels <mvels@google.com>:
Internal change
PiperOrigin-RevId: 415162872
--
64378549b110d5f5762185a5906c520fba70f0e7 by Abseil Team <absl-team@google.com>:
Fix a typo in the comments
PiperOrigin-RevId: 415088461
--
41aae8322e913b82710153c22b97c611fdb6e1fb by Abseil Team <absl-team@google.com>:
Switch from `connect` to `rt_sigreturn` -- the latter is much less problematic
for system call sandboxes.
PiperOrigin-RevId: 415073965
--
870c5e3388b6a35611bff538626fe7a1c8c87171 by Abseil Team <absl-team@google.com>:
Add ABSL_HAVE_HWADDRESS_SANITIZER and ABSL_HAVE_LEAK_SANITIZER
PiperOrigin-RevId: 414871189
--
f213ed60a66b58da7ac40555adfb1d529ff0a4db by Derek Mauro <dmauro@google.com>:
Remove reference to __SANITIZE_MEMORY__, which does not exist
It appears to have been copied by pattern matching from the ASAN/TSAN
code blocks.
|
3 years ago |
|
Abseil Team
|
3e1983c5c0 |
Export of internal Abseil changes
-- a9ea60e9c0ccd744b6f12fd021dbedfe826dfe84 by Matt Kulukundis <kfm@google.com>: Add an internal hook to allow keeping flags in sync with global state. Rollforward, except continue including hashtablez_flags.h in absl_flags.h so users don't break. PiperOrigin-RevId: 412198044 -- 183e5c440b68c797ce4a82102f94f41c97a14674 by Martijn Vels <mvels@google.com>: Internal cleanups and changes PiperOrigin-RevId: 412083793 -- 3740faf7c5a2e1723e3c7e4d1b3f3db7cbec6e61 by Abseil Team <absl-team@google.com>: Mark Cord::Clear() with the ABSL_ATTRIBUTE_REINITIALIZES attribute. This prevents false positives in the clang-tidy check bugprone-use-after-move; it allows Clear() to be called on a moved-from Cord without any warnings, and the Cord will thereafter be regarded as initialized again. PiperOrigin-RevId: 412082757 -- a730d3f4ba06b55ae50386920a0544592069ac01 by Abseil Team <absl-team@google.com>: StrJoin: Support iterators that do not have an `operator->` Allows using `StrJoin` with iterators that do not have an `operator->`. The `operator->` requirement for input iterators was dropped in C++20. PiperOrigin-RevId: 412066130 -- 6773c0ced2caa6a7855898298faecc584f3997ec by Andy Soffer <asoffer@google.com>: Rollback of internal hook for keeping flags in sync with global state. PiperOrigin-RevId: 411895027 -- 4e7016a2fb88ce97853ef85ad5b4f76998eacca1 by Matt Kulukundis <kfm@google.com>: Add an internal hook to allow keeping flags in sync with global state. PiperOrigin-RevId: 411867376 -- 2a7d4056e467b6b5d8a7aa9398d6cb5454c10fc5 by Martijn Vels <mvels@google.com>: Internal change PiperOrigin-RevId: 411806932 GitOrigin-RevId: a9ea60e9c0ccd744b6f12fd021dbedfe826dfe84 Change-Id: Ib35bb7b40774979ed2ad205bbb1744b1085eae78 |
3 years ago |
|
Abseil Team
|
72c7651111 |
Export of internal Abseil changes
-- 28ba85cbeb766341969fda677bbd8df8615b349f by Martijn Vels <mvels@google.com>: Internally allow for CordRepFlat sizes up to 256KB The default flat size will remain at 4KB, but this allows for future specializations such as file io and compression where larger buffers can be used providing 'zero copy' streaming optimizations. PiperOrigin-RevId: 411095538 -- 0c1cae5ff376a0955a5b67868723836665737801 by Abseil Team <absl-team@google.com>: cord_rep_btree: Reduce nesting in `DestroyTree` using guard clauses PiperOrigin-RevId: 411093741 GitOrigin-RevId: 28ba85cbeb766341969fda677bbd8df8615b349f Change-Id: Ieb84ed409f1e8a8bca74edebbf2d097a9ec828a7 |
3 years ago |
|
Abseil Team
|
39f46faa69 |
Export of internal Abseil changes
-- 83e4cdf03a4d702b30e69204060de09e462e23c6 by Greg Falcon <gfalcon@google.com>: Revert the crc addition to RefcountAndFlags, and restore related comments to their original state. In development, the implementation of SetExpectedCrc() changed, and there is no longer a need to track the CRC status in the refcount. Since the distinction between IsOne() and IsMutable() is subtle *and unused*, removing it now can help avoid subtle bugs in the future. This distinction can always be added back later, if it proves necessary. Keep the reserved bit for now; all it costs is one extra mask instruction in the refcount checks, and space for extra state in Cord is always hard to find. PiperOrigin-RevId: 408647038 -- ee67585cf66954176615271f50f8b278119dd138 by Greg Falcon <gfalcon@google.com>: Implement Cord::SetExpectedChecksum() and Cord::ExpectedChecksum(). SetExpectedChecksum() will store a uint32_t out-of-band alongside a Cord's data. This value persists through copies and assignments. Mutating operations on a Cord cause the value to be forgotten. ExpectedChecksum() retrieves the stored value, if present. This API is intended for storing a CRC32C checksum alongside data, allowing checksums to be passed through dataflows and validated at the final step. However, this API is agnostic to the meaning of the stored value. No CRC32C validation is performed by these new APIs. This implementation adds a new CordRep node, CordRepCrc. A CordRepCrc may (currently) only live at the top of a tree. This allows traversal logic to be agnostic to these nodes, instead putting the needed branches at the mutation level. This also implements the property requested from API review, that any mutation is guaranteed to permanently forget the stored CRC. PiperOrigin-RevId: 408611221 -- a86f592402b37c854ebdc77d2b9b425451a7a675 by Martijn Vels <mvels@google.com>: Move 'ExtractResult' into CordRep The result of an extract operation is logically identical for any tree implementation, and having a single type makes 'tree independent' implementation in cord.cc more concise. PiperOrigin-RevId: 408332408 -- baa7647e21db59a87f75af9cac62172ce38a0f71 by Abseil Team <absl-team@google.com>: Replace usages of `assert` macros with `ABSL_HARDENING_ASSERT`. PiperOrigin-RevId: 408272133 -- c7658133d8662c39fa5035fc93a364c7c3d327e0 by Martijn Vels <mvels@google.com>: Add CordRepBtree::ExtractAppendBuffer PiperOrigin-RevId: 407944179 -- 5775100363b5890ebfe710fadebf040445eab991 by Martijn Vels <mvels@google.com>: Add CordRepConcat::ExtractAppendBuffer PiperOrigin-RevId: 407932968 -- 9f520ba1600a93352c78f644a369c7c76195ee86 by Greg Falcon <gfalcon@google.com>: Add cordz tracking for crc nodes. This also adds a new kSetExpectedChecksum method to the list of tracked methods. This is presently unused but will be used soon. PiperOrigin-RevId: 407884120 GitOrigin-RevId: 83e4cdf03a4d702b30e69204060de09e462e23c6 Change-Id: I134ace2d87215813eaa60a282996a33884676c06 |
3 years ago |
|
Abseil Team
|
b6b62c72b4 |
Export of internal Abseil changes
-- f49e405201d2ffd5955503fa8ad0f08ec0cdfb2b by Martijn Vels <mvels@google.com>: Add common [container.requirements] type definitions to `CharRange` and `ChunkRange` The presence of these allow these range classes to be used in various utility functions which require some minimum type of container. For example, this change allows tests to use `EXPECT_THAT(cord.Chunks(), ElementsAre(...))` PiperOrigin-RevId: 406941278 -- 0c195f073632e21d9a4bce158047b2ba8551c2d1 by Evan Brown <ezb@google.com>: Use explicit exponential growth in SubstituteAndAppendArray. PiperOrigin-RevId: 406931952 -- afb043bccd809a55cab78abadb7548a057d9eda0 by Jorg Brown <jorg@google.com>: Use longer var names in macro to avoid clang-tidy warning PiperOrigin-RevId: 406930978 -- 80397e2604e6b3d929a34742c3a32581b34d3ac4 by Martijn Vels <mvels@google.com>: Add future kAppendBuffer and kPrependBuffer API trackers for Cordz sampling PiperOrigin-RevId: 406912759 -- e910ce919ef83933f08a690e8b7325c7cc5b6d5d by Martijn Vels <mvels@google.com>: Implement Prepend(string_view) in terms of PrependArray(string_view, MethodIdentifier). PiperOrigin-RevId: 406891665 -- c9cff43d4c0568ed01f2fca0f6ef038ae03112b5 by Martijn Vels <mvels@google.com>: Add 'Rebuild' logic to CordRepBtree There are btree hostile scenarios where an application could perform repeated split/insert/merge operations on a cord leading to a tree exceeding the maximum height. While this should be rare in practice, this change adds a Rebuild() method that will rebuild a tree with a 100% fill factor, and we will invoke this rebuild when a tree exceeds the maximum height. This basically follows the similar 'balance' logic in Concat trees (although the latter is common in Concat uses) PiperOrigin-RevId: 406875739 -- 5b2b8fb88f1ebfdc1c670088152da2cb2ea4c376 by Martijn Vels <mvels@google.com>: Add 'in place' enabled RemoveSuffix An in-place RemoveSuffix is more efficient than SubTree() as it can directly modify privately owned nodes and flats allowing easy re-use of free capacity in right-most flats that may turn into Substring edges when using SubTree. PiperOrigin-RevId: 406431230 -- f09903c0a3d7344f59aaf1380a16ea10829217d4 by Derek Mauro <dmauro@google.com>: Internal change PiperOrigin-RevId: 406430373 -- 9957af575c33bb18dc170572a4ee8cc5901df6b2 by Greg Falcon <gfalcon@google.com>: Initial groundwork to allow storing checksum data inside CordRep instances. This uses a RefcountAndFlags bit that was reserved for this purpose, and will be leveraged in a follow-up change to allow attaching checksums to a Cord's value. This change splits RefcountAndFlags::IsOne() into two distinct operations: * IsOne(): This returns true when the associated CordRep is not shared with other threads. This is useful for functions that consume CordRep instances; for example, code that consumes an unshared CordRep can assume ownership of its children without modifying those refcounts. * IsMutable(): This returns true when the associated CordRep reference is not shared with other threads, *and* does not store an associated checksum value. This is useful for functions that modify a CordRep's contents: code may modify the bytes of a mutable-unshared CordRep without fear of races with other threads, or of invalidating a stored checksum. The tricky part of this CL is ensuring that the correct choice between IsMutable() and IsOne() was made at each point. An incorrect application of IsOne() could lead to correctness bugs in the future. Code conditioned on IsOne() may delete the CordRep in question, or assume ownership of its children, but must not modify the CordRep's data without explicitly adjusting the CRC. PiperOrigin-RevId: 406191103 -- 686544814079e5ab6d4593cca0c068b510be400a by Martijn Vels <mvels@google.com>: Reduce the size in the LargeString test when running with Sanitizers PiperOrigin-RevId: 406186945 -- 735b4490bdb695c35731f06ce4b8de14ce2be6ed by Alex Strelnikov <strel@google.com>: Release absl::SimpleHexAtoi. PiperOrigin-RevId: 406143188 GitOrigin-RevId: f49e405201d2ffd5955503fa8ad0f08ec0cdfb2b Change-Id: Ic6527ac40fa03ea02ca813e8bb7868a219544de4 |
3 years ago |
|
Abseil Team
|
cfbf5bf948 |
Export of internal Abseil changes
-- 77e710b0ced5792a328e88bcb938a41484bf4cdc by Saleem Abdulrasool <abdulras@google.com>: absl: add an implementation for UnscaledCycleClock on RISCV Add an implementation for UnscaledCycleClock on RISC-V targets. PiperOrigin-RevId: 395982312 -- 84430fce6760c488ca36401cd530f44268ac710d by Martijn Vels <mvels@google.com>: Harden CordRepBtreeReader against reading up to or beyond EOF This change hardens the reader to Next() calls on EOF situations. It changes the 'consumed()' property inside CordRepBtreeReader into a 'remaining()' property which is easier to understand and use than the 'consumed()' property QED the function documentation and use in cord.cc This change also adds the CharIterator test to the CordTest fixture enabling them to be run with btree cords. PiperOrigin-RevId: 395971732 -- 6557e628f2613169da8f693189223acb30e07833 by Martijn Vels <mvels@google.com>: Add AdvanceAndRead() test addressing the edge case surfaced in b/197776822 This adds a test explicitly exercising all possible AdvanceAndRead() calls on CharIterator. As per the linked bug, a partial or full small read ending exactly at the end of the last edge of a btree cord results in an attempt to read beyond that last edge and subsequent failure. We will fix the bug and enable these tests for btree in a subsequent change. PiperOrigin-RevId: 395958317 GitOrigin-RevId: 77e710b0ced5792a328e88bcb938a41484bf4cdc Change-Id: Ie6e21ce36980515165af7cf046cf199ecbe0ddb0 |
3 years ago |
|
Abseil Team
|
4bb9e39c88 |
Export of internal Abseil changes
-- f73e17cb24f7878933fc100bd9bfc39fce190b64 by Derek Mauro <dmauro@google.com>: Internal change PiperOrigin-RevId: 394306402 -- 3d3eeffa4e37f63aa50fec1b90858043b40fe377 by Abseil Team <absl-team@google.com>: Release a few more absl::Cord unit tests that were accidentally omitted from the OSS release before. PiperOrigin-RevId: 394016464 -- 8a77a8eb93d021aadd8fdf43e219bf35328001ad by CJ Johnson <johnsoncj@google.com>: Fix typo in identifier PiperOrigin-RevId: 394000560 -- d87206c7c8e045b03d74b91e47ef3db0eb47a17b by Derek Mauro <dmauro@google.com>: Fix typo: RandenHwAes PiperOrigin-RevId: 393879427 -- 980a3402eea77b0c77fb20dd124203002ff791ba by Derek Mauro <dmauro@google.com>: Adds macros `ABSL_LTS_RELEASE_VERSION` and `ABSL_LTS_RELEASE_PATCH_LEVEL` to allow projects to detect if an LTS version is being used. Fixes #1006 PiperOrigin-RevId: 393807178 -- aecc7ed34de718c64733dab76621eacb5af9af5f by CJ Johnson <johnsoncj@google.com>: Change `alloc` to `allocator` to match the fact that other identifiers are full words PiperOrigin-RevId: 393794869 -- ad754bbcf7b78f5d51ed5f39193ac3159429b2b4 by Derek Mauro <dmauro@google.com>: Remove self-include of cord_rep_btree.h PiperOrigin-RevId: 393792085 -- f8e937a0d8fe26400560754f3596e3c21bb6d0d7 by Abseil Team <absl-team@google.com>: Fix trivial typo in comment. PiperOrigin-RevId: 393770527 -- 7a58ca5d708038d222c6a2b6ff5076b4ceffd370 by Tomas Dzetkulic <dzetkulic@google.com>: Update Cord::AppendArray resize policy. PiperOrigin-RevId: 393362184 -- 316050d171190d9d6312cadf88b1cc2db2d1caa7 by Abseil Team <absl-team@google.com>: Add a new top level profiling/ directory to the Abseil library PiperOrigin-RevId: 393358109 -- 0dbb8e10f7fa4a7ac74e12b178e936a67b266c51 by CJ Johnson <johnsoncj@google.com>: Switch to the more common enable_if pattern of ` = 0` in InlinedVector PiperOrigin-RevId: 393301549 -- 136d3068ce33b50ac820e8bd01395a7164d5181f by Abseil Team <absl-team@google.com>: Clean up typedefs in internal/inlined_vector.h PiperOrigin-RevId: 393181754 GitOrigin-RevId: f73e17cb24f7878933fc100bd9bfc39fce190b64 Change-Id: I0c4cd4d71d97bd1bf651701b6302ea3d9ac59b66 |
3 years ago |
|
Abseil Team
|
8e088c5f3c |
Export of internal Abseil changes
-- 77cd6291781bc39e8472c706163d6951fe2ae573 by Derek Mauro <dmauro@google.com>: absl::uint128: Use intrinsics for more operations when available This change also inlines the division and modulus operators when intrinsics are available for better code generation. Fixes #987 PiperOrigin-RevId: 389895706 -- fa23339584599e07ebcb4d0a857e2553b017757c by Abseil Team <absl-team@google.com>: only hide retired flags if human-readable output is requested PiperOrigin-RevId: 389835452 -- f1111f2b88359d4b253d4d81681c8a488458a36e by Martijn Vels <mvels@google.com>: Add helpers IsFlat(), IsExternal(), etc to improve readability PiperOrigin-RevId: 389779333 -- 785b8712261e41695ebeeb64b4317f93b37adc11 by Martijn Vels <mvels@google.com>: Split off 'concat' and 'btree' RepMemoryUsageLeaf and RepMemoryUsageDataEdge PiperOrigin-RevId: 389701120 -- 5264bffebffc2b377bf7e18f0ce69a3ed38c6629 by CJ Johnson <johnsoncj@google.com>: Eagerly destroy `Callback` in `absl::Cleanup` PiperOrigin-RevId: 389678813 -- a05312f0668458e97c50ca932c8f974c1508ebf2 by Abseil Team <absl-team@google.com>: Have one instance of empty_group per program, rather than one per translation unit. https://stackoverflow.com/questions/185624/static-variables-in-an-inlined-function PiperOrigin-RevId: 389185845 GitOrigin-RevId: 77cd6291781bc39e8472c706163d6951fe2ae573 Change-Id: Iac8d9cb27707a9562c831c77a552d1fb4bb0405f |
3 years ago |
|
Abseil Team
|
bf31a10b65 |
Export of internal Abseil changes
-- 93c607726d663800b4bfa472cba043fd3f5d0e97 by Derek Mauro <dmauro@google.com>: Internal change PiperOrigin-RevId: 389158822 -- 55b3bb50bbc168567c6ba25d07df2c2c39e864af by Martijn Vels <mvels@google.com>: Change CordRepRing alternative implementation to CordRepBtree alternative. This changes makes CordRepBtree (BTREE) the alternative to CordRepConcat (CONCAT) trees, enabled through the internal / experimental 'cord_btree_enabled' latch. PiperOrigin-RevId: 389030571 -- d6fc346143606c096bca8eb5029e4c429ac6e305 by Todd Lipcon <tlipcon@google.com>: Fix a small typo in SequenceLock doc comment PiperOrigin-RevId: 388972936 -- e46f9245dce8b4150e3ca2664e0cf42b75f90a83 by Martijn Vels <mvels@google.com>: Add 'shallow' validation mode to CordRepBtree which will be the default for internal assertions. PiperOrigin-RevId: 388753606 -- b5e74f163b490beb006f848ace67bb650433fe13 by Martijn Vels <mvels@google.com>: Add btree statistics to CordzInfo, and reduce rounding errors PiperOrigin-RevId: 388715878 -- 105bcbf80de649937e693b29b18220f9e6841a51 by Evan Brown <ezb@google.com>: Skip length checking when constructing absl::string_view from `const char*`. The length check causes unnecessary code bloat. PiperOrigin-RevId: 388271741 -- bed595158f24839efe49c65ae483f797d79fe0ae by Derek Mauro <dmauro@google.com>: Internal change PiperOrigin-RevId: 387713428 GitOrigin-RevId: 93c607726d663800b4bfa472cba043fd3f5d0e97 Change-Id: I2a4840f5ffcd7f70b7d7d45cce66f23c42cf565f |
3 years ago |
|
Abseil Team
|
89c531c1e0 |
Export of internal Abseil changes
-- e1a0989213908927f05002ab7697955ad7dc5632 by Martijn Vels <mvels@google.com>: Introduce CordRepBtreeReader CordRepBtreeReader provides forward navigation on cord btrees with absolute positional (offset) context, iterating over btree data in absl::string_view chunks. PiperOrigin-RevId: 387585161 -- 206d298e2bccb998731995cb05717b31fa9d90ec by Abseil Team <absl-team@google.com>: Internal change PiperOrigin-RevId: 387577465 -- f07fafe8a400a4f5dfef186d1a3b61fb7f709fe5 by Abseil Team <absl-team@google.com>: This change adds debug-build enforcement that the inputs to absl::c_set_intersection are sorted, which is a prerequisite of std::set_intersection and required for correct operation of the algorithm. PiperOrigin-RevId: 387446657 -- 2ca15c6361bb758be7fb88cae82bf8489b4d3364 by Abseil Team <absl-team@google.com>: Change BadStatusOrAccess::what() to contain status_.ToString() This ensures that on uncaught exception propagation that would cause program termination, the message contains information on the error which caused the failure. Lazy initialization of what_ is a value judgement: if most callers are expected to call status() not what(), lazy initialization is correct. If most callers are expected to call what(), it should be initialized on construction to avoid atomic operation overhead. PiperOrigin-RevId: 387402243 -- 3e855084e104dc972a0c4385395e6d8e8465127f by Gennadiy Rozental <rogeeff@google.com>: LSC: Standardize access to GoogleTest flags on GTEST_FLAG_GET/GTEST_FLAG_SET This change is necessary to move Googletest flags out of the testing:: namespace without breaking code. These new macros will continue to be required for code that needs to work both inside Google's monorepo and outside in OSS, but can be used anywhere inside the monorepo. PiperOrigin-RevId: 387396025 -- 1ccf5895a15059ef689af5c4817d7b84f73190be by Gennadiy Rozental <rogeeff@google.com>: Import of CCTZ from GitHub. PiperOrigin-RevId: 387388496 GitOrigin-RevId: e1a0989213908927f05002ab7697955ad7dc5632 Change-Id: I3606d9ce29d909a3555e662e9df564202cf5068d |
3 years ago |
|
Abseil Team
|
5dd2407243 |
Export of internal Abseil changes
-- 60b8e77be4bab1bbd3b4c3b70054879229634511 by Derek Mauro <dmauro@google.com>: Use _MSVC_LANG for some C++ dialect checks since MSVC doesn't set __cplusplus accurately by default. https://devblogs.microsoft.com/cppblog/msvc-now-correctly-reports-__cplusplus/ See GitHub #722. PiperOrigin-RevId: 371362181 -- 5d736accdff04db0e722f377c0d79f2d3ed53263 by Martijn Vels <mvels@google.com>: Fix the estimated memory size for CordRepExternal PiperOrigin-RevId: 371350380 -- eaaa1d8a167aeca67a2aa3a098a2b61a9d72172f by Martijn Vels <mvels@google.com>: Remove flakes by not enforcing re-allocated pointers do never match original Tests that do multiple updates could end up with the original allocated pointer on a 2nd resize, so the 'EqIfPrivate' should not assume that if we do 'not' have the capacity that all following relocations will never match the original. We only care about 'pointer unchanged if private and there is capacity', trying to establish 'pointer changed at some point due to re-allocation; is pointless. PiperOrigin-RevId: 371338965 -- d1837bee6bade1902b095c1cbf64231668bb84c5 by Martijn Vels <mvels@google.com>: Undo inline of small data copy in cord This leads to a performance regression as the code is not inlined (absent hard FDO inputs), and there are no suitable tail call options. PiperOrigin-RevId: 371332332 -- 06dc64b833069efc7d18b11df607c8c22be690da by Martijn Vels <mvels@google.com>: Add final instrumentation for Cordz and remove 'old' cordz logic. This change instruments the last cord function for cordz. It removes the 'old' functions: set_tree, replace_tree, UpdateCordzStatistics and RecordMetrics. PiperOrigin-RevId: 371219909 -- a5e0be538579c603052feec03e6d9910c43ea787 by Martijn Vels <mvels@google.com>: Extend the life of CordRep* if inside a snapshot If a snapshot (potentially) includes the current CordzInfo, we need to extent the lifetime of the CordRep*, as the snapshot 'point in time' observation of the cord should ideally be preserved. PiperOrigin-RevId: 371146151 -- 74d77a89774cd6c8ecdeebee0193b294a39383d6 by Martijn Vels <mvels@google.com>: Instrument std::string consuming methods: ctor, operator=, Append and Prepend This change moves the 'steal into CordRep' logic into a separate function so we can use it directly in the ctor, operator assign and append and prepend, allowing Cordz instrumentation with the proper method attributes. The assign operator is implemented in AssignLargeString leaving the dispatch inlined in cord.h (which as a side effects also allows clean tail calls in the AssignLargeString method) PiperOrigin-RevId: 371094756 -- b39effc45266b7ce2e7f96caa3b16cb6e3acc2dd by Martijn Vels <mvels@google.com>: Add Cordz instrumentation to CordReader PiperOrigin-RevId: 370990181 GitOrigin-RevId: 60b8e77be4bab1bbd3b4c3b70054879229634511 Change-Id: I96af62e6f1a643e8b1228ae01e6c84e33706bb05 |
4 years ago |
|
Abseil Team
|
d96e287417 |
Export of internal Abseil changes
--
f825cf3feb6db06522b2b4ee785de7dfa325780d by Martijn Vels <mvels@google.com>:
Move Cordz test helpers to cordz_test_helpers library
PiperOrigin-RevId: 370059941
--
5080249da6a4f5cc2b546aed48503fd028670379 by Martijn Vels <mvels@google.com>:
Add new Cordz instrumentation on AppendTree.
PiperOrigin-RevId: 369968167
--
21092b889fad34ec605894e311b436d5f417456f by Benjamin Barenblat <bbaren@google.com>:
Round floats using round(x), not static_cast<int>(x + 0.5)
Adding 0.5 to an IEEE float may cause one bit of precision loss, which
is enough to change the result in certain cases. For example,
static_cast<int>(std::round(0.49999999999999994)) == 0
static_cast<int>(0.49999999999999994 + 0.5) == 1
PiperOrigin-RevId: 369926519
GitOrigin-RevId: f825cf3feb6db06522b2b4ee785de7dfa325780d
Change-Id: Ib78ce1faec79f06578933db5dc6fc05de043ead1
|
4 years ago |
|
Abseil Team
|
2faed9dd2a |
Export of internal Abseil changes
-- 6cfac39ea1266f01f195de5eb4c9a6fc9ea9b20a by Matt Kulukundis <kfm@google.com>: Fix a typo PiperOrigin-RevId: 366174890 -- 6ee8c58647aef171d394e59fa06f9bf8cd0306ec by Derek Mauro <dmauro@google.com>: Adds `ABSL_ATTRIBUTE_LIFETIME_BOUND` and applies it to the `const std::string&` constructor of `absl::string_view`. Compilers that support this attribute will emit a warning if the parameter does not have sufficient lifetime. PiperOrigin-RevId: 366027738 -- b944427d96e4b436b8fa0fe396c2a1118dbbbd13 by Abseil Team <absl-team@google.com>: Calls to `ResetToEmpty()` from `ClearSlow` use ~3% of the time in assignment. However, `ClearSlow()` is only used in contexts where `data_` is immediately reassigned. Rename `ClearSlow()` into `UnrefTree()` and remove `data_` resetting. PiperOrigin-RevId: 365977213 -- 7428b3147a5672c8bb55649efa3a1cfe19b52a8b by Abseil Team <absl-team@google.com>: Fix CordRepRing diabolical growth The 'Mutable' function in CordRepRing was over-eager in doubling capacity, which lead to 'ludicrous' growth in the diabolical test case as added to cord_test. This CL fixes the doubling for growing shared reps, and tempers CordRepRing growth for non shared capacity to 1.5 instead of 2, which is more inline with a conservative growth we also have in tree cord. After this change, CordRepRing no longer swamps the heap into the shadow realm, and is in effect reducing the memory used compared to the tree implementation. With a diabolical 5000 bytes growth pattern: Tree cord: 1523520 bytes Ring cord: 274232 bytes PiperOrigin-RevId: 365915857 -- f24d4aee48b03c5a7980664df903f947cbb198e8 by Andy Getzendanner <durandal@google.com>: Import of CCTZ from GitHub. PiperOrigin-RevId: 365873932 GitOrigin-RevId: 6cfac39ea1266f01f195de5eb4c9a6fc9ea9b20a Change-Id: I9737aa215ac732c9785a1d0032c77aba62330f12 |
4 years ago |
|
Abseil Team
|
9fe3519549 |
Export of internal Abseil changes
-- 6b5be2524a088d0f4e8475794dc71232a24e94d8 by Abseil Team <absl-team@google.com>: Enable ABSL_HAVE_ATTRIBUTE_WEAK for Windows with Clang >= 9.0.0 The bug (https://bugs.llvm.org/show_bug.cgi?id=37598) motivated the workaround was fixed in 9.0.0. PiperOrigin-RevId: 365682074 -- c16b7784978a370658dce6d82cb7055316a79bcc by Abseil Team <absl-team@google.com>: Add IsFlat() evaluation to GetFlatAux for RingBuffer PiperOrigin-RevId: 365666501 -- c064eb686a3c036e093e71126c45f97d3a921569 by Abseil Team <absl-team@google.com>: Implement C++11 compatible std::remove_cvref added in C++20 PiperOrigin-RevId: 365606639 -- af2e7e055172da914e63c05308aedb68e197661e by Abseil Team <absl-team@google.com>: Add IsFlat() support to CordRepRing PiperOrigin-RevId: 365562090 -- 2cfeff9280f4967c4f828812bfe153b4e9cbabb7 by Abseil Team <absl-team@google.com>: Make unit test for TryFlat on 'substring of rep' explicit PiperOrigin-RevId: 365081382 GitOrigin-RevId: 6b5be2524a088d0f4e8475794dc71232a24e94d8 Change-Id: Ibb577748176217ce237614a6fe77c05375a97003 |
4 years ago |
|
Abseil Team
|
9c6a50fdd8 |
Export of internal Abseil changes
-- 4ff721439234e91caf6f7b772e5f554e7dd423c8 by Benjamin Barenblat <bbaren@google.com>: Remove endian-sensitivity from hash slow path Prior to this commit, the Abseil hash fast path was endian-agnostic, but the slow path assumed a little-endian platform. Change the slow path to be endian-correct, ensuring that values produced by the fast and slow paths are equal even on big-endian systems. PiperOrigin-RevId: 355424258 -- 7f4fe1aa4de46ad0a2ef19fa9c061fc12a7391ed by Abseil Team <absl-team@google.com>: Directly store CordzInfo in the InlineData data contents of InlineRep This greatly reduces the cost of coping and moving cords. Especially the move constructor and move assignment are now back to lean loads and stores without needing any CordzInfo lookups for tracked cords. PiperOrigin-RevId: 355409161 -- 3ca4ca84ed6d98f1e383ffd8d12c28876e905bb3 by Abseil Team <absl-team@google.com>: Add #include <unordered_map> PiperOrigin-RevId: 355386114 -- 30b0ffad0621971b3135148fcc9e183b0dd2a6bb by Abseil Team <absl-team@google.com>: Optimize Cord copy constructor This change avoids double stores of the Cord copy constructor from the zero init of the InlineData / InlineRep contents followed by the assignment and inlines the copy constructor. PiperOrigin-RevId: 355287939 -- 0c043fa7b6e41ca7cefc5edc1e17ad46223e4e77 by CJ Johnson <johnsoncj@google.com>: Now that the absl::Cleanup example returns absl::Status, since we decided on absl::FailedPreconditionError, the precondition should be a positive statement and then the check should be failure to adhere to that positive statement PiperOrigin-RevId: 355216923 -- 9ed922ca5d28fe8790ec6bc0837cf39fbcc92896 by Gennadiy Rozental <rogeeff@google.com>: Do not set mvsc linker flags for clang-cl (fixes #874) Import of https://github.com/abseil/abseil-cpp/pull/891 PiperOrigin-RevId: 355199380 GitOrigin-RevId: 4ff721439234e91caf6f7b772e5f554e7dd423c8 Change-Id: I3d9d2383549720d7a91f9108dfcd979ad6632fce |
4 years ago |
|
Abseil Team
|
a9a4956020 |
Export of internal Abseil changes
-- c68f1886f5e8fd90eb0c2d2e68feaf00a7cdacda by CJ Johnson <johnsoncj@google.com>: Introduce absl::Cleanup to the OSS repo PiperOrigin-RevId: 354583156 -- 17030cf388e10f7eb959e3e566326d1072ce392e by Abseil Team <absl-team@google.com>: Internal change only PiperOrigin-RevId: 354574953 -- e979d7236d4f3252e79ddda6739b67a9a326bf6d by CJ Johnson <johnsoncj@google.com>: Internal change PiperOrigin-RevId: 354545297 -- 7ea02b3783f7f49ef97d86a8f6580a19cc57df14 by Abseil Team <absl-team@google.com>: Pre-allocate memory for vectors where the size is known. PiperOrigin-RevId: 354344576 -- 9246c7cb11f1d6444f79ebe25acc69a8a9b870e0 by Matt Kulukundis <kfm@google.com>: Add support for Elbrus 2000 (e2k) Import of https://github.com/abseil/abseil-cpp/pull/889 PiperOrigin-RevId: 354344013 -- 0fc93d359cc1fb307552e917b37b7b2e7eed822f by Abseil Team <absl-team@google.com>: Integrate CordRepRing logic into cord (but do not enable it) PiperOrigin-RevId: 354312238 -- eda05622f7da71466723acb33403f783529df24b by Abseil Team <absl-team@google.com>: Protect ignore diagnostic with "__has_warning". PiperOrigin-RevId: 354112334 -- 47716c5d8fb10efa4fdd801d28bac414c6f8ec32 by Abseil Team <absl-team@google.com>: Rearrange InlinedVector copy constructor and destructor to treat a few special cases inline and then tail-call a non-inlined routine for the rest. In particular, we optimize for empty vectors in both cases. Added a couple of benchmarks that copy either an InlVec<int64> or an InlVec<InlVec<int64>>. Speed difference: ``` BM_CopyTrivial/0 0.92ns +- 0% 0.47ns +- 0% -48.91% (p=0.000 n=11+12) BM_CopyTrivial/1 0.92ns +- 0% 1.15ns +- 0% +25.00% (p=0.000 n=10+9) BM_CopyTrivial/8 8.57ns +- 0% 10.72ns +- 1% +25.16% (p=0.000 n=10+12) BM_CopyNonTrivial/0 3.21ns +- 0% 0.70ns +- 0% -78.23% (p=0.000 n=12+10) BM_CopyNonTrivial/1 5.88ns +- 1% 5.51ns +- 0% -6.28% (p=0.000 n=10+8) BM_CopyNonTrivial/8 21.5ns +- 1% 15.2ns +- 2% -29.23% (p=0.000 n=12+12) ``` Note: the slowdowns are a few cycles which is expected given the procedure call added in that case. We decided this is a good tradeoff given the code size reductions and the more significant speedups for empty vectors. Size difference (as measured by nm): ``` BM_CopyTrivial from 1048 bytes to 326 bytes. BM_CopyNonTrivial from 749 bytes to 470 bytes. ``` Code size for a large binary drops by ~500KB (from 349415719 to 348906015 348906191). All of the benchmarks that showed a significant difference: Ones that improve with this CL: ``` BM_CopyNonTrivial/0 3.21ns +- 0% 0.70ns +- 0% -78.23% (p=0.000 n=12+10) BM_InlinedVectorFillString/0 0.93ns +- 0% 0.24ns +- 0% -74.19% (p=0.000 n=12+10) BM_InlinedVectorAssignments/1 10.5ns +- 0% 4.1ns +- 0% -60.64% (p=0.000 n=11+10) BM_InlinedVectorAssignments/2 10.7ns +- 0% 4.4ns +- 0% -59.08% (p=0.000 n=11+11) BM_CopyTrivial/0 0.92ns +- 0% 0.47ns +- 0% -48.91% (p=0.000 n=11+12) BM_CopyNonTrivial/8 21.5ns +- 1% 15.2ns +- 2% -29.23% (p=0.000 n=12+12) BM_StdVectorEmpty 0.47ns +- 1% 0.35ns +- 0% -24.73% (p=0.000 n=12+12) BM_StdVectorSize 0.46ns +- 2% 0.35ns +- 0% -24.32% (p=0.000 n=12+12) BM_SwapElements<LargeCopyableOnly>/0 3.44ns +- 0% 2.76ns +- 1% -19.83% (p=0.000 n=11+11) BM_InlinedVectorFillRange/256 20.7ns +- 1% 17.8ns +- 0% -14.08% (p=0.000 n=12+9) BM_CopyNonTrivial/1 5.88ns +- 1% 5.51ns +- 0% -6.28% (p=0.000 n=10+8) BM_SwapElements<LargeCopyableMovable>/1 4.19ns +- 0% 3.95ns +- 1% -5.63% (p=0.000 n=11+12) BM_SwapElements<LargeCopyableMovableSwappable>/1 4.18ns +- 0% 3.99ns +- 0% -4.70% (p=0.000 n=9+11) BM_SwapElements<LargeCopyableMovable>/0 2.41ns +- 0% 2.31ns +- 0% -4.45% (p=0.000 n=12+12) BM_InlinedVectorFillRange/64 8.25ns +- 0% 8.04ns +- 0% -2.51% (p=0.000 n=12+11) BM_SwapElements<LargeCopyableOnly>/1 82.4ns +- 0% 81.5ns +- 0% -1.06% (p=0.000 n=12+12) ``` Ones that get worse with this CL: ``` BM_CopyTrivial/1 0.92ns +- 0% 1.15ns +- 0% +25.00% (p=0.000 n=10+9) BM_CopyTrivial/8 8.57ns +- 0% 10.72ns +- 1% +25.16% (p=0.000 n=10+12) BM_SwapElements<LargeCopyableMovableSwappable>/512 1.48ns +- 1% 1.66ns +- 1% +11.88% (p=0.000 n=12+12) BM_InlinedVectorFillString/1 11.5ns +- 0% 12.8ns +- 1% +11.62% (p=0.000 n=12+11) BM_SwapElements<LargeCopyableMovableSwappable>/64 1.48ns +- 2% 1.66ns +- 1% +11.66% (p=0.000 n=12+11) BM_SwapElements<LargeCopyableMovableSwappable>/1k 1.48ns +- 1% 1.65ns +- 2% +11.32% (p=0.000 n=12+12) BM_SwapElements<LargeCopyableMovable>/512 1.48ns +- 2% 1.58ns +- 4% +6.62% (p=0.000 n=11+12) BM_SwapElements<LargeCopyableMovable>/1k 1.49ns +- 2% 1.58ns +- 3% +6.05% (p=0.000 n=12+12) BM_SwapElements<LargeCopyableMovable>/64 1.48ns +- 2% 1.57ns +- 4% +6.04% (p=0.000 n=11+12) BM_InlinedVectorFillRange/1 4.81ns +- 0% 5.05ns +- 0% +4.83% (p=0.000 n=11+11) BM_InlinedVectorFillString/8 79.4ns +- 1% 83.1ns +- 1% +4.64% (p=0.000 n=10+12) BM_StdVectorFillString/1 16.3ns +- 0% 16.6ns +- 0% +2.13% (p=0.000 n=11+8) ``` PiperOrigin-RevId: 353906786 -- 8e26518b3cec9c598e5e9573c46c3bd1b03a67ef by Abseil Team <absl-team@google.com>: Internal change PiperOrigin-RevId: 353737330 -- f206ae0983e58c9904ed8b8f05f9caf564a446be by Matt Kulukundis <kfm@google.com>: Import of CCTZ from GitHub. PiperOrigin-RevId: 353682256 GitOrigin-RevId: c68f1886f5e8fd90eb0c2d2e68feaf00a7cdacda Change-Id: I5790c1036c4f543c701d1039848fabf7ae881ad8 |
4 years ago |
|
Abseil Team
|
4b915e7092 |
Export of internal Abseil changes
-- 77c85460dc3c46593b231c5161ac55273bb0c7ef by Abseil Team <absl-team@google.com>: Support int128 in SimpleAtoi. PiperOrigin-RevId: 337946262 -- 0be53049ccf8309650e4e22f23b290e5f75ee828 by Gennadiy Rozental <rogeeff@google.com>: Update build scripts to use --mount instead of --volume to mount the host data into a container. This is somewhat more verbose, but more readable and flexible. More importantly this is what docker documentation is recomending to use now. PiperOrigin-RevId: 337898761 -- 3bc877f1679fdf61ecbf4365287a0403cfc9b53e by Samuel Benzaquen <sbenza@google.com>: Add Cord constructor for constinit instances. PiperOrigin-RevId: 337871148 -- 8b87701892b9c325e78ad4e8e4f16b785a744622 by Chris Kennelly <ckennelly@google.com>: Use the address of kSeed, rather than its value. We initialize kSeed with &kSeed, so these are equivalent, but kSeed requires a load from memory while &kSeed can be formed as a RIP-relative lea. PiperOrigin-RevId: 337868415 GitOrigin-RevId: 77c85460dc3c46593b231c5161ac55273bb0c7ef Change-Id: I3d06c18a123f1be29dad5801e8625952dc41cd95 |
4 years ago |
|
Abseil Team
|
4b2fbb4adb |
Export of internal Abseil changes
-- a5af5874c1c5cc02bd2a748d455321f82b6f2a93 by Andy Getzendanner <durandal@google.com>: fix compile fails with asan and -Wredundant-decls Import of https://github.com/abseil/abseil-cpp/pull/801 PiperOrigin-RevId: 336693223 -- ed9df42ab2b742386c6692c2bed015374c919d9c by Derek Mauro <dmauro@google.com>: Fix integer conversion warning Fixes #814 PiperOrigin-RevId: 336651814 -- 0ab4c23884e72dce17b67c1eb520f9dbb802565d by Derek Mauro <dmauro@google.com>: Internal change PiperOrigin-RevId: 336585378 -- eba0e3dccd52a6e91bcff84075bef0affc650b74 by Matt Kulukundis <kfm@google.com>: Add bitset operations to Futex helper. PiperOrigin-RevId: 336409368 -- 8b0709a8b4500bf5f0af4b602d76a298d81645e8 by Abseil Team <absl-team@google.com>: Fix code indentation in a comment. PiperOrigin-RevId: 336368167 -- bc3961c87a7e7760c10319a5b0349c279f7ae3ad by Samuel Benzaquen <sbenza@google.com>: Improve performance of the registry: - Reduce contention - Reduce memory usage for each flag by `6*sizeof(void*)`. - Replace one immortal allocation per-flag with a single one for all the flags - Slightly improve single-threaded performance by avoiding the std::map indirections. PiperOrigin-RevId: 336365904 -- 264ad9f28f935aad8b6b1437f8bf804fa9104346 by Abseil Team <absl-team@google.com>: Fix typo in comment on absl::Condition. PiperOrigin-RevId: 336311680 -- b5b808a8c75ca0df7b09eff9a423ec171d80f771 by Derek Mauro <dmauro@google.com>: Add missing Apache license headers PiperOrigin-RevId: 336294980 -- 89446c3a4793df8b95060385cf3e219357c3db1d by Andy Soffer <asoffer@google.com>: Internal changes PiperOrigin-RevId: 336287465 -- 57c8be4e294881bc79a6a44b8e4bf7ecbb19b9b9 by Matt Kulukundis <kfm@google.com>: Extract Futex from an implementation detail of Wait to a private interface. PiperOrigin-RevId: 336123209 GitOrigin-RevId: a5af5874c1c5cc02bd2a748d455321f82b6f2a93 Change-Id: Ie5a0ebe28e571814e3e11d4c05ca308523ccf311 |
4 years ago |
|
Abseil Team
|
768eb2ca28 |
Export of internal Abseil changes
-- f012012ef78234a6a4585321b67d7b7c92ebc266 by Laramie Leavitt <lar@google.com>: Slight restructuring of absl/random/internal randen implementation. Convert round-keys.inc into randen_round_keys.cc file. Consistently use a 128-bit pointer type for internal method parameters. This allows simpler pointer arithmetic in C++ & permits removal of some constants and casts. Remove some redundancy in comments & constexpr variables. Specifically, all references to Randen algorithm parameters use RandenTraits; duplication in RandenSlow removed. PiperOrigin-RevId: 312190313 -- dc8b42e054046741e9ed65335bfdface997c6063 by Abseil Team <absl-team@google.com>: Internal change. PiperOrigin-RevId: 312167304 -- f13d248fafaf206492c1362c3574031aea3abaf7 by Matthew Brown <matthewbr@google.com>: Cleanup StrFormat extensions a little. PiperOrigin-RevId: 312166336 -- 9d9117589667afe2332bb7ad42bc967ca7c54502 by Derek Mauro <dmauro@google.com>: Internal change PiperOrigin-RevId: 312105213 -- 9a12b9b3aa0e59b8ee6cf9408ed0029045543a9b by Abseil Team <absl-team@google.com>: Complete IGNORE_TYPE macro renaming. PiperOrigin-RevId: 311999699 -- 64756f20d61021d999bd0d4c15e9ad3857382f57 by Gennadiy Rozental <rogeeff@google.com>: Switch to fixed bytes specific default value. This fixes the Abseil Flags for big endian platforms. PiperOrigin-RevId: 311844448 -- bdbe6b5b29791dbc3816ada1828458b3010ff1e9 by Laramie Leavitt <lar@google.com>: Change many distribution tests to use pcg_engine as a deterministic source of entropy. It's reasonable to test that the BitGen itself has good entropy, however when testing the cross product of all random distributions x all the architecture variations x all submitted changes results in a large number of tests. In order to account for these failures while still using good entropy requires that our allowed sigma need to account for all of these independent tests. Our current sigma values are too restrictive, and we see a lot of failures, so we have to either relax the sigma values or convert some of the statistical tests to use deterministic values. This changelist does the latter. PiperOrigin-RevId: 311840096 GitOrigin-RevId: f012012ef78234a6a4585321b67d7b7c92ebc266 Change-Id: Ic84886f38ff30d7d72c126e9b63c9a61eb729a1a |
5 years ago |