Limited to architectures where it is actually needed by gcc for any
calls to stdatomic.h atomic_is_lock_free for now.
We should also add it to other architectures too, along with lockful
atomic r/m/w operations for sizes that can't be handled natively, but
that's a lot more work. It is also necessary for -fno-inline-atomics
but we're missing a lot of other symbols for that too, to be fixed.
For now, this should enable the OpenSSL build to complete on these
architectures again after I reverted a local change.
XXX pullup-10
Changes in 1.2.13 (13 Oct 2022)
- Fix configure issue that discarded provided CC definition
- Correct incorrect inputs provided to the CRC functions
- Repair prototypes and exporting of new CRC functions
- Fix inflateBack to detect invalid input with distances too far
- Have infback() deliver all of the available output up to any error
- Fix a bug when getting a gzip header extra field with inflate()
- Fix bug in block type selection when Z_FIXED used
- Tighten deflateBound bounds
- Remove deleted assembler code references
- Various portability and appearance improvements
Changes in 1.2.12 (27 Mar 2022)
- Cygwin does not have _wopen(), so do not create gzopen_w() there
- Permit a deflateParams() parameter change as soon as possible
- Limit hash table inserts after switch from stored deflate
- Fix bug when window full in deflate_stored()
- Fix CLEAR_HASH macro to be usable as a single statement
- Avoid a conversion error in gzseek when off_t type too small
- Have Makefile return non-zero error code on test failure
- Avoid some conversion warnings in gzread.c and gzwrite.c
- Update use of errno for newer Windows CE versions
- Small speedup to inflate [psumbera]
- Return an error if the gzputs string length can't fit in an int
- Add address checking in clang to -w option of configure
- Don't compute check value for raw inflate if asked to validate
- Handle case where inflateSync used when header never processed
- Avoid the use of ptrdiff_t
- Avoid an undefined behavior of memcpy() in gzappend()
- Avoid undefined behaviors of memcpy() in gz*printf()
- Avoid an undefined behavior of memcpy() in _tr_stored_block()
- Make the names in functions declarations identical to definitions
- Remove old assembler code in which bugs have manifested
- Fix deflateEnd() to not report an error at start of raw deflate
- Add legal disclaimer to README
- Emphasize the need to continue decompressing gzip members
- Correct the initialization requirements for deflateInit2()
- Fix a bug that can crash deflate on some input when using Z_FIXED
- Assure that the number of bits for deflatePrime() is valid
- Use a structure to make globals in enough.c evident
- Use a macro for the printf format of big_t in enough.c
- Clean up code style in enough.c, update version
- Use inline function instead of macro for index in enough.c
- Clarify that prefix codes are counted in enough.c
- Show all the codes for the maximum tables size in enough.c
- Add gznorm.c example, which normalizes gzip files
- Fix the zran.c example to work on a multiple-member gzip file
- Add tables for crc32_combine(), to speed it up by a factor of 200
- Add crc32_combine_gen() and crc32_combine_op() for fast combines
- Speed up software CRC-32 computation by a factor of 1.5 to 3
- Use atomic test and set, if available, for dynamic CRC tables
- Don't bother computing check value after successful inflateSync()
- Correct comment in crc32.c
- Add use of the ARMv8 crc32 instructions when requested
- Use ARM crc32 instructions if the ARM architecture has them
- Explicitly note that the 32-bit check values are 32 bits
- Avoid adding empty gzip member after gzflush with Z_FINISH
- Fix memory leak on error in gzlog.c
- Fix error in comment on the polynomial representation of a byte
- Clarify gz* function interfaces, referring to parameter names
- Change macro name in inflate.c to avoid collision in VxWorks
- Correct typo in blast.c
- Improve portability of contrib/minizip
- Fix indentation in minizip's zip.c
- Replace black/white with allow/block. (theresa-m)
- minizip warning fix if MAXU32 already defined. (gvollant)
- Fix unztell64() in minizip to work past 4GB. (Daniël Hörchner)
- Clean up minizip to reduce warnings for testing
- Add fallthrough comments for gcc
- Eliminate use of ULL constants
- Separate out address sanitizing from warnings in configure
- Remove destructive aspects of make distclean
- Check for cc masquerading as gcc or clang in configure
- Fix crc32.c to compile local functions only if used
Changes in 1.2.11 (15 Jan 2017)
- Fix deflate stored bug when pulling last block from window
- Permit immediate deflateParams changes before any deflate input
Changes in 1.2.13 (13 Oct 2022)
- Fix configure issue that discarded provided CC definition
- Correct incorrect inputs provided to the CRC functions
- Repair prototypes and exporting of new CRC functions
- Fix inflateBack to detect invalid input with distances too far
- Have infback() deliver all of the available output up to any error
- Fix a bug when getting a gzip header extra field with inflate()
- Fix bug in block type selection when Z_FIXED used
- Tighten deflateBound bounds
- Remove deleted assembler code references
- Various portability and appearance improvements
Changes in 1.2.12 (27 Mar 2022)
- Cygwin does not have _wopen(), so do not create gzopen_w() there
- Permit a deflateParams() parameter change as soon as possible
- Limit hash table inserts after switch from stored deflate
- Fix bug when window full in deflate_stored()
- Fix CLEAR_HASH macro to be usable as a single statement
- Avoid a conversion error in gzseek when off_t type too small
- Have Makefile return non-zero error code on test failure
- Avoid some conversion warnings in gzread.c and gzwrite.c
- Update use of errno for newer Windows CE versions
- Small speedup to inflate [psumbera]
- Return an error if the gzputs string length can't fit in an int
- Add address checking in clang to -w option of configure
- Don't compute check value for raw inflate if asked to validate
- Handle case where inflateSync used when header never processed
- Avoid the use of ptrdiff_t
- Avoid an undefined behavior of memcpy() in gzappend()
- Avoid undefined behaviors of memcpy() in gz*printf()
- Avoid an undefined behavior of memcpy() in _tr_stored_block()
- Make the names in functions declarations identical to definitions
- Remove old assembler code in which bugs have manifested
- Fix deflateEnd() to not report an error at start of raw deflate
- Add legal disclaimer to README
- Emphasize the need to continue decompressing gzip members
- Correct the initialization requirements for deflateInit2()
- Fix a bug that can crash deflate on some input when using Z_FIXED
- Assure that the number of bits for deflatePrime() is valid
- Use a structure to make globals in enough.c evident
- Use a macro for the printf format of big_t in enough.c
- Clean up code style in enough.c, update version
- Use inline function instead of macro for index in enough.c
- Clarify that prefix codes are counted in enough.c
- Show all the codes for the maximum tables size in enough.c
- Add gznorm.c example, which normalizes gzip files
- Fix the zran.c example to work on a multiple-member gzip file
- Add tables for crc32_combine(), to speed it up by a factor of 200
- Add crc32_combine_gen() and crc32_combine_op() for fast combines
- Speed up software CRC-32 computation by a factor of 1.5 to 3
- Use atomic test and set, if available, for dynamic CRC tables
- Don't bother computing check value after successful inflateSync()
- Correct comment in crc32.c
- Add use of the ARMv8 crc32 instructions when requested
- Use ARM crc32 instructions if the ARM architecture has them
- Explicitly note that the 32-bit check values are 32 bits
- Avoid adding empty gzip member after gzflush with Z_FINISH
- Fix memory leak on error in gzlog.c
- Fix error in comment on the polynomial representation of a byte
- Clarify gz* function interfaces, referring to parameter names
- Change macro name in inflate.c to avoid collision in VxWorks
- Correct typo in blast.c
- Improve portability of contrib/minizip
- Fix indentation in minizip's zip.c
- Replace black/white with allow/block. (theresa-m)
- minizip warning fix if MAXU32 already defined. (gvollant)
- Fix unztell64() in minizip to work past 4GB. (Daniël Hörchner)
- Clean up minizip to reduce warnings for testing
- Add fallthrough comments for gcc
- Eliminate use of ULL constants
- Separate out address sanitizing from warnings in configure
- Remove destructive aspects of make distclean
- Check for cc masquerading as gcc or clang in configure
- Fix crc32.c to compile local functions only if used
Changes in 1.2.11 (15 Jan 2017)
- Fix deflate stored bug when pulling last block from window
- Permit immediate deflateParams changes before any deflate input
gcc generates calls to this symbol in programs that use
__sync_*_compare_and_swap, which require full sequential consistency
barriers, including store-before-load ordering on both sides of the
atomic; none of the release/acquire operations guarantee that, so we
have to insert explicit DMB instructions.
Note: gcc's own definition omits some of the DMB instructions, but I
can't prove that it's correct that way -- stores preceding the CAS
must complete before the load part of the CAS, and the store part of
the CAS must complete before loads following the CAS. Maybe there's
some way to prove that one of these orderings is guaranteed some
other way than a DMB but I'm not seeing it, and store-before-load
ordering is hard to understand.
Patch by skrll@ based on a patch by mrg@, soliloquy in commit message
by me.
The more-compatible LOCK ADD $0,-N(%rsp) turns out to be cheaper
than MFENCE anyway. Let's save some space and maintenance and rip
out the hotpatching for it.
- membar_enter/exit ordering was backwards.
- membar_enter doesn't make any sense for load anyway.
- Switch to membar_release for store and membar_acquire for load.
The only sensible orderings for a simple load or store are acquire or
release, respectively, or sequential consistency. This never
provided correct sequential consistency before -- we should really
make it conditional on memmodel but I don't know offhand what the
values of memmodel might be and this is at least better than before.
The names membar_enter/exit were unclear, and the documentation of
membar_enter has disagreed with the implementations on sparc,
powerpc, and even x86(!) for the entire time it has been in NetBSD.
The terms `acquire' and `release' are ubiquitous in the literature
today, and have been adopted in the C and C++ standards to mean
load-before-load/store and load/store-before-store, respectively,
which are exactly the orderings required by acquiring and releasing a
mutex, as well as other useful applications like decrementing a
reference count and then freeing the underlying object if it went to
zero.
Originally I proposed changing one word in the documentation for
membar_enter to make it load-before-load/store instead of
store-before-load/store, i.e., to make it an acquire barrier. I
proposed this on the grounds that
(a) all implementations guarantee load-before-load/store,
(b) some implementations fail to guarantee store-before-load/store,
and
(c) all uses in-tree assume load-before-load/store.
I verified parts (a) and (b) (except, for (a), powerpc didn't even
guarantee load-before-load/store -- isync isn't necessarily enough;
need lwsync in general -- but it _almost_ did, and it certainly didn't
guarantee store-before-load/store).
Part (c) might not be correct, however: under the mistaken assumption
that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then
membar-r/rw, I only audited the cases of membar_enter that _aren't_
immediately after an atomic-r/m/w. All of those cases assume
load-before-load/store. But my assumption was wrong -- there are
cases of atomic-r/m/w then membar-w/rw that would be broken by
changing to atomic-r/m/w then membar-r/rw:
https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html
Furthermore, the name membar_enter has been adopted in other places
like OpenBSD where it actually does follow the documentation and
guarantee store-before-load/store, even if that order is not useful.
So the name membar_enter currently lives in a bad place where it
means either of two things -- r/rw or w/rw.
With this change, we deprecate membar_enter/exit, introduce
membar_acquire/release as better names for the useful pair (r/rw and
rw/w), and make sure the implementation of membar_enter guarantees
both what was documented _and_ what was implemented, making it an
alias for membar_sync.
While here, rework all of the membar_* definitions and aliases. The
new logic follows a rule to make it easier to audit:
membar_X is defined as an alias for membar_Y iff membar_X is
guaranteed by membar_Y.
The `no stronger than' relation is (the transitive closure of):
- membar_consumer (r/r) is guaranteed by membar_acquire (r/rw)
- membar_producer (w/w) is guaranteed by membar_release (rw/w)
- membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw)
- membar_release (rw/w) is guaranteed by membar_sync (rw/rw)
And, for the deprecated membars:
- membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by
membar_sync (rw/rw)
- membar_exit (rw/w) is guaranteed by membar_release (rw/w)
(membar_exit is identical to membar_release, but the name is
deprecated.)
Finally, while here, annotate some of the instructions with their
semantics. For powerpc, leave an essay with citations on the
unfortunate but -- as far as I can tell -- necessary decision to use
lwsync, not isync, for membar_acquire and membar_consumer.
Also add membar(3) and atomic(3) man page links.
This will be deprecated soon but let's avoid leaving rakes to trip on
with it arising from disagreement over the documentation (W/RW) and
implementation and usage (R/RW).
This will be deprecated soon but let's avoid leaving rakes to trip on
with it arising from disagreement over the documentation (W/RW) and
implementation and usage (R/RW).
This will be deprecated soon but let's avoid leaving rakes to trip on
with it arising from disagreement over the documentation (W/RW) and
implementation and usage (R/RW).
This will be deprecated soon but let's avoid leaving rakes to trip on
with it arising from disagreement over the documentation (W/RW) and
implementation and usage (R/RW).
This will be deprecated soon but let's avoid leaving rakes to trip on
with it arising from disagreement over the documentation (W/RW) and
implementation and usage (R/RW).
On x86, every store is a store-release, so there is no need for any
barrier. But this wasn't a barrier anyway; it was just a store,
which was redundant with the store of the return address to the stack
implied by CALL even if issuing a store made a difference.
lfence is only needed for MD logic, such as operations on I/O memory
rather than normal cacheable memory, or special instructions like
RDTSC -- never for MI synchronization between threads/CPUs. No need
for hot-patching to do lfence here.
(The x86_lfence function might reasonably be patched on i386 to do
lfence for MD logic, but it isn't now and this doesn't change that.)
In TSO this is the only memory barrier ever needed, and somehow we
got this wrong and instead issued an unnecessary membar #LoadLoad --
not needed even in PSO let alone in TSO.
XXX Apparently we may run userland programs with PSO or RMO, in which
case all of these membars need fixing:
PSO RMO
membar_consumer nop membar #LoadLoad
membar_producer membar #StoreStore membar #StoreStore
membar_enter nop membar #LoadLoad|LoadStore
membar_exit membar #StoreStore membar #LoadStore|StoreStore
membar_sync membar #StoreLoad|StoreStore
membar #...everything...
But at least this fixes the TSO case in which we run the kernel.
Also I'm not sure there's any non-TSO hardware out there in practice.
skrll