1. Set the capacity to 1 when IsExclusive is true.
2. Set it to 3 because Try shows we never have more the 3 elements when IsExclusive is false.
MozReview-Commit-ID: dmIeE6ZTeh
--HG--
extra : rebase_source : fc6b173d3a389d11424238f76cbae1f496155ba5
extra : source : c6203c27b8020999ed39ff43ddf5173f5111ef6b
This improves performance because we won't try to fix the permissions of the file every time the nsIHandlerService implementation wants to open it, and makes it easier to access the file path from xpcshell tests.
MozReview-Commit-ID: DZmLa92qUnZ
--HG--
extra : source : bd0fdce03cc7442079e96318d3a9b519a4d403a9
PLDHashTable takes the result of the hash function and multiplies it by
kGoldenRatio to ensure that it has a good distribution of bits across
the 32-bit hash value, and then zeroes out the low bit so that it can be
used for the collision flag. This result is called hash0. From hash0
it computes two different numbers used to find entries in the table
storage: hash1 is used to find an initial position in the table to
begin searching for an entry; hash2 is then used to repeatedly offset
that position (mod the size of the table) to build a chain of positions
to search.
In a table with capacity 2^c entries, hash1 is simply the upper c bits
of hash0. This patch does not change this.
Prior to this patch, hash2 was the c bits below hash1, padded at the low
end with zeroes when c > 16. (Note that bug 927705, changeset
1a02bec165e16f370cace3da21bb2b377a0a7242, increased the maximum capacity
from 2^23 to 2^26 since 2^23 was sometimes insufficient!) This manner
of computing hash2 is problematic because it increases the risk of long
chains for very large tables, since there is less variation in the hash2
result due to the zero padding.
So this patch changes the hash2 computation by using the low bits of
hash0 instead of shifting it around, thus avoiding 0 bits in parts of
the hash2 value that are significant.
Note that this changes what hash2 is in all cases except when the table
capacity is exactly 2^16, so it does change our hashing characteristics.
For tables with capacity less than 2^16, it should be using a different
second hash, but with the same amount of random-ish data. For tables
with capacity greater than 2^16, it should be using more random-ish
data.
MozReview-Commit-ID: JvnxAMBY711
--HG--
extra : transplant_source : %8A%25%FB%E3H%B8_%F1G%F6%3E%0B%29%DF%20%FF%D8%E1%AEw
PLDHashTable's entry store has two types of unoccupied entries: free
entries and removed entries. The search of a chain of entries
(determined by the hash value) in the entry store to search for an entry
can stop at free entries, but it continues across removed entries,
because removed entries are entries that may have been skipped over when
we were adding the value we're searching for to the hash, but have since
been removed. For live entries, we also maintain this distinction by
using one bit of storage for a collision flag, which notes that if the
hashtable entry is removed, its place in the entry store must become a
removed entry rather than a free entry.
When we add a new entry to the table, Add's semantics require that we
return an existing entry if there is one, and only create a new entry if
no existing entry exists. (Bug 1352198 suggests the possibility of a
faster alternative Add API where the caller guarantees that the key is
not already in the hashtable.) When we search for the existing entry,
we must thus continue the search across removed entries, even though we
record the first removed entry found to return if the search for an
existing entry fails.
The existing code adds the collision flag through the entire table
search during an Add. This patch changes that behavior so that we only
add the collision flag prior to finding the first removed entry. Adding
it after we find the first removed entry is unnecessary, since we are
not making that entry part of a path to a new entry. If it is part of a
path to an existing entry, it will already have the collision flag set.
This patch effectively puts an if (!firstRemoved) around the else branch
of the if (MOZ_UNLIKELY(EntryIsRemoved(entry))), and then refactors that
condition outwards since it is now around the contents of both the if
and else branches.
MozReview-Commit-ID: CsXnMYttHVy
--HG--
extra : transplant_source : 0T%B0%FA%C0%85v%8B%16%E7%81%03p%F5K%97%B1%9E%92%27
This uses std::atomic rather than mozilla::Atomic since mozilla::Atomic
does not support using different memory synchronization for different
atomic operations on the same variable.
The added comments could use careful review since, while they reflect my
understanding of the issue, I don't consider myself an expert on the
topic.
MozReview-Commit-ID: 7xByCXt17Dr
--HG--
extra : transplant_source : 8%8Ci%CC%EA%0F%CF%C7%3E%F1%93%F5%C9%ED9%84%F9%3Evx
PLDHashTable's entry store has two types of unoccupied entries: free
entries and removed entries. The search of a chain of entries
(determined by the hash value) in the entry store to search for an entry
can stop at free entries, but it continues across removed entries,
because removed entries are entries that may have been skipped over when
we were adding the value we're searching for to the hash, but have since
been removed. For live entries, we also maintain this distinction by
using one bit of storage for a collision flag, which notes that if the
hashtable entry is removed, its place in the entry store must become a
removed entry rather than a free entry.
When we add a new entry to the table, Add's semantics require that we
return an existing entry if there is one, and only create a new entry if
no existing entry exists. (Bug 1352198 suggests the possibility of a
faster alternative Add API where the caller guarantees that the key is
not already in the hashtable.) When we search for the existing entry,
we must thus continue the search across removed entries, even though we
record the first removed entry found to return if the search for an
existing entry fails.
The existing code adds the collision flag through the entire table
search during an Add. This patch changes that behavior so that we only
add the collision flag prior to finding the first removed entry. Adding
it after we find the first removed entry is unnecessary, since we are
not making that entry part of a path to a new entry. If it is part of a
path to an existing entry, it will already have the collision flag set.
This patch effectively puts an if (!firstRemoved) around the else branch
of the if (MOZ_UNLIKELY(EntryIsRemoved(entry))), and then refactors that
condition outwards since it is now around the contents of both the if
and else branches.
MozReview-Commit-ID: CsXnMYttHVy
--HG--
extra : transplant_source : W4%B8%BA%D5p%102%1B%8D%83%23%E0s%B3%B0f%0D%05%AE
This is the same optimization made for ThreadSafeAutoRefCnt in bug
1277709. However, nsStringBuffer uses a 32-bit reference count all the
time, so it can't easily use ThreadSafeAutoRefCnt.
MozReview-Commit-ID: LpB3xaYbaEE
--HG--
extra : transplant_source : %89%FA%02%402%B1%83%17%81%09%EEw%B0%85%A0%B2%8DK/%D6
This uses std::atomic rather than mozilla::Atomic since mozilla::Atomic
does not support using different memory synchronization for different
atomic operations on the same variable.
The added comments could use careful review since, while they reflect my
understanding of the issue, I don't consider myself an expert on the
topic.
MozReview-Commit-ID: 7xByCXt17Dr
--HG--
extra : transplant_source : %8DM%88%E8%B7%B4%D8a%D6%F5%3F%9B%DC%09X%F3%7C%98%DE%21
For testing purposes it will be useful to be able to trigger crashes in Rust
code. Being able to trigger a panic seems like a good place to start. This
will make it easier to validate improvements in crash reporting.
MozReview-Commit-ID: Bh5rBieLLWW
--HG--
rename : toolkit/crashreporter/test/unit/test_crash_moz_crash.js => toolkit/crashreporter/test/unit/test_crash_rust_panic.js
extra : rebase_source : ba9f626ca2d2852f966e93401b8f8f7543615310
extra : source : 8c9117de1e7f40af42b7cbce25bc3780c032fe45
This adds an arena allocator that can be used as a drop-in replacement for
NSPR's PLArena. Example usage for defining an 8-byte aligned allocator that
uses a 4K arena size:
mozilla::ArenaAllocator<4096,8> a;
void* memory = a.Allocate(200);
For testing purposes it will be useful to be able to trigger crashes in Rust
code. Being able to trigger a panic seems like a good place to start. This
will make it easier to validate improvements in crash reporting.
MozReview-Commit-ID: Bh5rBieLLWW
--HG--
rename : toolkit/crashreporter/test/unit/test_crash_moz_crash.js => toolkit/crashreporter/test/unit/test_crash_rust_panic.js
extra : rebase_source : 0ac96469629a7a933fcf3bf6720c448db45543eb
extra : source : 8c9117de1e7f40af42b7cbce25bc3780c032fe45
It's a very general mechanism for replacing the implementation of
printf_stderr().
It's primarily used by the profiler, sparingly, and not in an important way.
Worse, it prevents us from using MOZ_LOG in the profiler, which is something I
want. Because if any code that locks gPSMutex also calls MOZ_LOG, that then
calls printf_stderr(), which calls profiler_log(), which locks gPSMutex, which
deadlocks.
The only other use of set_stderr_callback() is for the ultra-hacky,
for-local-use-only copy_stderr_to_file() function, which was added for B2G
debugging and is no longer necessary.
This patch removes set_stderr_callback() altogether.
--HG--
extra : rebase_source : d31ecb482fe5899f62dc56a38e87d91f9271bab0
This tracks TlsAlloc() and TlsFree() calls on Windows for diagnosing crashes when a proces reaches
its limit (1088) for TLS slots. Tracking of TLS allocation is done by intercepting TlsAlloc() and
TlsFree() in kernel32.dll. After initialization, we start tracking the number of allocated TLS
slots. If the number of observed TLS allocations exceeds a high water mark, we record the stack
when TlsAlloc() is called, and the recorded stacks gets serialized in a JSON string ready for
crash annotation.
MozReview-Commit-ID: 5fHVr0eiMy5
This patch changes the crashreporter client code as well as the crash service
code to compute a SHA256 hash of a crash' minidump file and add it to the
crash ping. The crash service code computes the hash on the fly before handing
over the crash to the crash manager; the crash manager will then add it to the
crash ping. The crashreporter client on the other hand sends the hash via the
ping it generates but it also adds it to the event file so that the crash
manager can pick it up and send it along with its own crash ping. On Fennec
the crashreporter activity takes care of computing the hash.
SHA256 hash computation uses nsICryptoHash in the crash service, the
java.security.MessageDigest class in Fennec, the bundled NSS library in the
crashreporter when running on Windows and Mac and the system-provided NSS
library under Linux. The latter is required because the crashreporter client
uses the system curl library which is linked to NSS and which would thus clash
with the bundled one if used together.
This patch introduces two new methods for the nsICrashService interface:
|getMinidumpForID()| and |getExtraFileForID()|, these reliably retrieve the
.dmp and .extra files associated with a crash and ensure the files exist
before returning. These new methods are used in the CrashService for
processing and will become the only way to reliably retrieve those files
from a crash ID.
MozReview-Commit-ID: 8BKvqj6URcO
--HG--
extra : source : a4d8291c56fcde00238ab3166bbe6af6dd602340
This change moves us away from NSPR primitives for our primary
synchronization primitives. We're still using PRMonitor for
ReentrantMonitor, however.
The benefits of this change:
* Slightly faster, as we don't have to deal with some of NSPR's overhead;
* Smaller datatypes. On POSIX platforms in particular, PRLock is
enormous. PRCondVar also has some unnecessary overhead.
* Less dynamic memory allocation. Out of necessity, Mutex and CondVar
allocated the NSPR data structures they needed, which lead to
unnecessary checks for failure.
While sizeof(Mutex) and sizeof(CondVar) may get bigger, since they're
embedding structures now, the total memory usage should be less.
* Less NSPR usage. This shouldn't need any explanation.
Carsten "Tomcat" Book
JW Wang
Paolo Amadini
L. David Baron
Olli Pettay
Bevis Tseng
Eric Rahm
Ted Mielczarek
Henri Sivonen
Iris Hsiao
Andrew McCreight
Jon Coppeard
Wes Kocher
Bobby Holley
Andrea Marchesini
emilio+bugs@crisal.io
Michael Layzell
Valentin Gosu
Thinker K.F. Li
Nicholas Nethercote
Cervantes Yu
Gabriele Svelto
Boris Zbarsky
Nathan Froyd
Ehsan Akhgari