- 04 5月, 2012 9 次提交
-
-
由 Rich Felker 提交于
this could lead to spurious failures of wide printf functions
-
由 Rich Felker 提交于
-
由 Rich Felker 提交于
musl does not support legacy 32-bit-off_t whatsoever. off_t is always 64 bit, and correct programs that use off_t and the standard functions will just work out of the box. (on glibc, they would require -D_FILE_OFFSET_BITS=64 to work.) however, some programs instead define _LARGEFILE64_SOURCE and use alternate versions of all the standard types and functions with "64" appended to their names. we do not want code to actually get linked against these functions (it's ugly and inconsistent), so macros are used instead of prototypes with weak aliases in the library itself. eventually the weak aliases may be added at the library level for the sake of using code that was originally built against glibc, but the macros will still be the desired solution in the headers.
-
由 Rich Felker 提交于
-
由 Rich Felker 提交于
these actually work, but for now they prohibit actually setting priority levels and report min/max priority as 0.
-
由 Rich Felker 提交于
-
由 Rich Felker 提交于
in6_* is in the reserved namespace, so this is valid
-
由 Rich Felker 提交于
pthread structure has been adjusted to match the glibc/GCC abi for where the canary is stored on i386 and x86_64. it will need variants for other archs to provide the added security of the canary's entropy, but even without that it still works as well as the old "minimal" ssp support. eventually such changes will be made anyway, since they are also needed for GCC/C11 thread-local storage support (not yet implemented). care is taken not to attempt initializing the thread pointer unless the program actually uses SSP (by reference to __stack_chk_fail).
-
由 Rich Felker 提交于
hopefully the annoyance of this will be minimal. these files all define internal interfaces which can change at any time; if different modules are using different versions of the interfaces, the library will badly break. ideally we would scan and add the dependency only for C files that actually reference the affected interfaces, but for now, err on the side of caution and force a rebuild of everything if any of them have changed. this commit is in preparation for the upcoming ssp overhaul commit, which will change internals of the pthread struct.
-
- 03 5月, 2012 6 次提交
-
-
由 Rich Felker 提交于
looks like nik copied these "extra arguments" from the i386 code. they're not actually arguments there, just 1-byte instructions to make sure the stack is aligned to 16 bytes after all the other arguments are pushed. since each push is 8 bytes on x86_64, they happened to have no effect here, but their presence is confusing and a minor waste of space.
-
由 Rich Felker 提交于
-
由 Rich Felker 提交于
-
由 Rich Felker 提交于
-
由 Rich Felker 提交于
it does not work; after further consideration, a separate Scrt1.s for pie really is essential. it would be nice if the unified approach worked, but the linker fails to generate the correct PLT entries and instead puts textrels in the main program, which don't work because the kernel maps the text read-only. new Scrt1.s will be committed soon in place of this.
-
由 Rich Felker 提交于
these are POSIX 2008 (previously GNU extension) functions that are rarely used. apparently they had never been tested before, since the end-of-string logic was completely missing. mbsnrtowcs is used by modern versions of bash for its glob implementation, and and this bug was causing tab completion to hang in an infinite loop.
-
- 02 5月, 2012 3 次提交
-
-
由 Rich Felker 提交于
these were at best of limited usefulness (for bootstrapping new systems, mainly) and at worst caused real kernel headers to get overwritten when upgrading libc. in case they're needed by anyone, the exact same files are now available in a new git repository: git://git.etalabs.net/mini-lkh
-
由 Rich Felker 提交于
the major change here is that CFLAGS is now a variable that can be changed entirely under user control, without causing essential flags to be lost. previously, "CFLAGS += ..." was valid in config.mak, but using "CFLAGS = ..." in config.mak would have badly broken the build process unless the user took care to copy the necessary flags out of the main Makefile. I have also added a distclean target that removes config.mak.
-
由 Rich Felker 提交于
as far as I can tell, it's not useful and never way. I wrote it way back under the assumption that non-weak symbols in the POSIX or extension namespace could conflict with legitimate uses of the same symbol name in the main program or other libraries, but that does not seem to be the case.
-
- 01 5月, 2012 2 次提交
-
-
由 Rich Felker 提交于
-
由 Rich Felker 提交于
-
- 30 4月, 2012 7 次提交
-
-
由 Rich Felker 提交于
this is a nonstandard function so it's not clear what conditions it should satisfy. my intent is that it be fast and exact for positive integral exponents when the result fits in the destination type, and fast and correctly rounded for small negative integral exponents. otherwise we aim for at most 1ulp error; it seems to differ from pow by at most 1ulp and it's often 2-5 times faster than pow.
-
由 Rich Felker 提交于
-
由 Rich Felker 提交于
this caused misreading of certain floating point values that are exact multiples of large powers of ten, unpredictable depending on prior stack contents.
-
由 Rich Felker 提交于
-
由 Rich Felker 提交于
untested
-
由 Rich Felker 提交于
-
由 Rich Felker 提交于
-
- 29 4月, 2012 2 次提交
-
-
由 Rich Felker 提交于
-
由 Rich Felker 提交于
unlike the old one, this one's algorithm does not suffer from potential stack overflow issues or pathologically bad performance on certain patterns. instead of backtracking, it uses a matching algorithm which I have not seen before (unsure whether I invented or re-invented it) that runs in O(1) space and O(nm) time. it may be possible to improve the time to O(n), but not without significantly greater complexity.
-
- 27 4月, 2012 2 次提交
-
-
由 Rich Felker 提交于
if the compiler provides a value, use it; otherwise fallback to the platform default (2).
-
由 Rich Felker 提交于
an invalid bracket expression must be treated as if the opening bracket were just a literal character. this is to fix a bug whereby POSIX left the behavior of the "[" shell command undefined due to it being an invalid bracket expression.
-
- 26 4月, 2012 1 次提交
-
-
由 Rich Felker 提交于
-
- 25 4月, 2012 6 次提交
-
-
由 Rich Felker 提交于
provide the minimal level of dynamic linker-to-debugger glue needed to let gdb find loaded libraries and load their symbols.
-
由 Rich Felker 提交于
the code is written to pre-init the thread pointer in static linked programs that pull in __stack_chk_fail or dynamic-linked programs that lookup the symbol. no explicit canary is set; the canary will be whatever happens to be in the thread structure at the offset gcc hard-coded. this can be improved later.
-
由 Rich Felker 提交于
otherwise this BADLY breaks if -funsigned-char is passed to gcc
-
由 Rich Felker 提交于
-
由 Rich Felker 提交于
i did some testing trying to switch malloc to use the new internal lock with priority inheritance, and my malloc contention test got 20-100 times slower. if priority inheritance futexes are this slow, it's simply too high a price to pay for avoiding priority inversion. maybe we can consider them somewhere down the road once the kernel folks get their act together on this (and perferably don't link it to glibc's inefficient lock API)... as such, i've switch __lock to use malloc's implementation of lightweight locks, and updated all the users of the code to use an array with a waiter count for their locks. this should give optimal performance in the vast majority of cases, and it's simple. malloc is still using its own internal copy of the lock code because it seems to yield measurably better performance with -O3 when it's inlined (20% or more difference in the contention stress test).
-
由 Rich Felker 提交于
this bug probably would have gone unnoticed since it's only used in the fallback code for systems where priority-inheritance locking fails. unfortunately this approach results in one spurious wake syscall on the final unlock, when there are no waiters remaining. the alternative (possibly better) would be to use broadcast wakes instead of reflagging the waiter unconditionally, and let each waiter reflag itself; this saves one syscall at the expense of invoking the "thundering herd" effect (worse performance degredation) when there are many waiters. ideally we would be able to update all of our locks to use an array of two ints rather than a single int, and use a separate counter system like proper mutexes use; then we could avoid all spurious wake calls without resorting to broadcasts. however, it's not clear to me that priority inheritance futexes support this usage. the kernel sets the waiters flag for them (just like we're doing now) and i can't tell if it's safe to bypass the kernel when unlocking just because we know (from private data, the waiter count) that there are no waiters. this is something that could be explored in the future.
-
- 24 4月, 2012 2 次提交
-
-
由 Rich Felker 提交于
we use priority inheritance futexes if possible so that the library cannot hit internal priority inversion deadlocks in the presence of realtime priority scheduling (full support to be added later).
-
由 Rich Felker 提交于
i tried to go with improving the old binary-search-based algorithm, but between growth in the number of ranges, bad performance, and lack of confidence in the binary search code's stability under changes in the table, i decided it was worth the extra 1.8k to have something clean and maintainable. also note that, like the alpha and punct tables, there's definitely room to optimize the nonspacing/wide tables by overlapping subtables. this is not a high priority, but i've begun looking into how to do it, and i suspect the table sizes can be roughly halved. if that turns out to be true, the new, fast, table-based implementation will be roughly the same size as if i had just extended the old binary search one.
-