| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
Codepaths streamlined. Improved performance for nonthreaded apps
when linked with a thread-enabled libc.
Minor iconv bug and some locale/thread related startup issues fixed.
These showed up in getting a gcj-compiled java helloworld app running.
Removed some old extension functions... _stdio_fdout and _stdio_fsfopen.
|
|
Fix (hopefully) a potential problem with failed freopen() calls. The
fix isn't tested since I've been working on the replacement stdio
core code which will go in after the next release.
|
|
|
|
Added a list of popen()'d to store pids and use waitpid() in pclose().
Loop on waitpid() failure due to EINTR as required.
Close parent's popen()'d FILEs in the {v}fork()'d child.
Fix failure exit code for failed execve().
|
|
were including libc-lock.h which had a bunch of weak pragmas. Also,
uClibc supplied a number of no-op weak thread functions even though
many weren't needed. This combined result was that sometimes the
functional versions of thread functions in pthread would not override
the weaks in libc.
While fixing this, I also prepended double-underscore to all necessary
weak thread funcs in uClibc, and removed all unused weaks.
I did a test build, but haven't tested this since these changes are
a backport from my working tree. I did test the changes there and
no longer need to explicitly add -lpthread in the perl build for
perl to pass its thread self tests.
|
|
|
|
Indirectly detected by gmp-4.1.2 self-tests and reported by
"Peter S. Mazinger" <ps.m@gmx.net>.
|
|
rm.patch:
* Define $(RM) as rm -f in Rules.mak and test/Rules.mak
(this is the same definition as gmake uses by default).
* Change all occurrences of rm and rm -f into $(RM).
|
|
wchar support is enabled.
|
|
encountering EOF changed with Defect Report #141. In the current
standard, the stream's EOF indicator is "sticky". Once it is set,
all further input from the stream should fail until the application
explicitly clears the EOF indicator (clearerr(), file positioning),
even if more data becomes available.
Fixed a bug in fgets. Wasn't checking for read errors.
Minor thread locking optimizations to avoid some unnecessary locking.
Remove the explicit calls to __builtin_* funcs, as we really need to
implement a more general solution.
|
|
I got my mind fixed in one mode and didn't comply with the standards.
Things should be fixed now, but comparision testing is difficult when
glibc's scanf is broken and they stubbornly refuse to even acknowledge
that it is... even when confronted by specific examples from the C99
standards and from an official C standard defect report.
|
|
for environments where long and long long are the same size.
|
|
Printf still needs some performance work done though.
Also some bufferless stdio cleanups.
|
|
configured with non-buffered stdio and non-wchar mode.
Fix a couple of bugs that showed up in minimalist configurations.
Update old_vfprintf to handle size qualifiers on the integer and %n
conversions. It now passed the glibc printf tests with the exception
of a floating point rounding difference.
|
|
|
|
|
|
|
|
even when just starting a new mb char.
wscanf would incorrectly unget in certain situations.
|
|
Should be standards compliant and with several optional features,
including support for hexadecimal float notation, locale awareness,
glibc-like locale-specific digit grouping with the `'' flag, and
positional arg support. I tested it pretty well (finding several
bugs in glibc's scanf in the process), but it is brand new so be
aware.
The *wprintf functions now support floating point output. Also, a
couple of bugs were squashed. Finally, %a/%A conversions are
now implemented.
Implement the glibc xlocale interface for thread-specific locale
support. Also add the various *_l(args, locale_t loc_arg) funcs.
NOTE!!! setlocale() is NOT threadsafe! NOTE!!!
The strto{floating point} conversion functions are now locale aware.
The also now support hexadecimal floating point notation.
Add the wcsto{floating point} conversion functions.
Fix a bug in mktime() related to dst. Note that unlike glibc's mktime,
uClibc's version always normalizes the struct tm before attempting
to determine the correct dst setting if tm_isdst == -1 on entry.
Add a stub version of the libintl functions. (untested)
Fixed a known memory leak in setlocale() related to the collation data.
Add lots of new config options (which Erik agreed to sort out :-),
including finally exposing some of the stripped down stdio configs.
Be careful with those though, as they haven't been tested in a
long time.
(temporary) GOTCHAs...
The ctype functions are currently incorrect for 8-bit locales. They
will be fixed shortly.
The ctype functions are now table-based, resulting in larger staticly
linked binaries. I'll be adding an option to use the old approach
in the stub locale configuration.
|
|
not setting the FILE bufread member to flag the end of the buffer.
Also, do not set bufgetc member if getc macro support is disabled.
|
|
|
|
|
|
to match that of current glibc; i.e. don't do anything and return success.
Apparently, php calls fflush() on a file opened as readonly before trying
to read. Eventually I'll add some config options to flag this and several
other instances of nonportable code.
|
|
back. Unfortunately, none of my test cases caught it.
|
|
%c, %s, and %[ specifiers. Note that scanf is undergoing rewrite so I
didn't bother optimizing this. I did run all my regression tests though.
Set EOF correctly for fmemopen on readonly streams. I really need to
check what glibc behavior is for the various open modes though.
|
|
is built without buffer support.
|
|
seeks to the end of the stream when append mode is set and we are
transitioning to write mode, so that subsequent ftell() return
values are correct.
Also fix _stdio_fopen to support fdopen() with append specified when
the underlying file didn't have O_APPEND set. It now sets the
O_APPEND flag as recommended by SUSv3 and is done by glibc.
|
|
|
|
Also added outdigit support and (legacy) YESSTR/NOSTR support.
|
|
instead of -1 as expected.
|
|
|
|
Hopefully locale support will build when cross compiling now. Collation is
still not supported, but that's what I'm currently working on. In the
next couple of days, I'll probably put up a couple of files for download
that will save people the trouble of generating all the glibc locales.
Added *wprintf functions, although they currently don't support floating
point. That will be fixed when I rewrite _dtostr... or possibly before.
Added the wcsto{inttype} functions.
Added iconv() and a mini iconv utility. The require locale support and
only provide for conversions involving the various unicode encodings
{ UCS-4*, UCS-2*, UTF-32*, UTF-16*, UTF-8 }, the 8-bit codesets built
with the locale data, and the internal WCHAR_T.
|
|
comment as an mjn3 only item.
|
|
Require printf format strings to be valid multibyte strings beginning and
ending in their initial shift state, as per the stds.
Fixed a bug in _wchar_wcsntoutf8s(). Don't store wcs position if dst is NULL.
Also, introduce an awful hack into _wchar_wcsntoutf8s() and wcsrtombs() in
order to support %ls in printf. See comments below for details.
Change behaviour of wc<->mb functions when in the C locale. Now they do
a 1-1 map for the range 0x80-UCHAR_MAX. This is for backwards compatibility
and consistency with the stds requirements that a printf format string by
a valid multibyte string beginning and ending in it's initial shift state.
|
|
floating point output, so at least implement that. But grouping will
have to wait for the rewrite of _dtostr.
|
|
when the grouping flag "'" is specified. Grouping for floating point values
may wait until I do a rewrite of the floating pt to string code...
|
|
been working on a new config system on and off for about 6 months
now, but I've never been fully satisfied. Well, I'm finally am
happy with the new config system, so here it is. This completely
removes the old uClibc configuration system, and replaces it with
an entirely new system based on LinuxKernelConf, from
http://www.xs4all.nl/~zippel/lc/
As it turns out, Linus has just merged LinuxKernelConf into Linux
2.5.45, so it looks like I made the right choice.
I have thus far updated only x86. I'll be updating the other
architectures shortly.
-Erik
|
|
-Erik
|
|
without custom stream support. In that case, it is necessary to do
a va_copy. Note: The affected code is not built in the stock config.
Also, make sure each va_copy has a matching va_end, as required by C99.
|
|
case.
|
|
the cookie pointer was set to NULL as a debugging aid. However,
some of the perl 5.8 test rely on being able to close stderr and
still try writing to it. So now, the cookie pointer and handler
function pointers are set to that it is a "normal" file with a
file descriptor of -1.
Note: The cookie pointer is reset to NULL if the FILE struct is
free'd by fclose.
|
|
sense to exclude it, gcc always supports long long, and we never
fully excluded long long anyways (off64_t for example).
-Erik
|
|
One involving %o and one involving %f.
|
|
I think they're fixed now (I've run a few tests).
Note: __fsetlocking() is not threadsafe... but glibc's doesn't appear to
be either.
|
|
needlessly. To do so increases the generated code size with bcc.
Eliminate duplicate define warnings in wstring.c.
Fix potentially broken preprocessor comparisons. The preprocessor
converts integers to maximal signed type, so inequality comparisons
involving UINTMAX_MAX, ULLONG_MAX, and (if no long long) ULONG_MAX
were potentially broken.
|
|
using
a 3.0.4 version of the sh-linux-gcc compiler.
|
|
This falls under undefined behavior wrt ANSI/ISO C99, but
SUSv3 seems to treat it as a no-op and it occurs in some apps.
Fixed a problem with _stdio_fwrite() not checking for underlying
write() failures.
Fixed both _stdio_fwrite() and _stdio_fread() to make sure that
the putc and getc macros were disabled if the stream was in
and error state.
The above changes should take care of a problem initially reported
by "Steven J. Hill" <sjhill@realitydiluted.com>.
|
|
|
