Age | Commit message (Collapse) | Author |
|
|
|
|
|
Found by the compiler, not me:
lib/lib.c:1053:30: warning: 'st2.st_dev' may be used uninitialized in this function [-Wmaybe-uninitialized]
if (st1.st_dev != st2.st_dev || st1.st_ino != st2.st_ino) continue;
~~~^~~~~~~
|
|
If we have a 15-byte name, we don't know whether comm actually matches
or is a truncated form of a longer name that has a common prefix.
For example, with "this-is-a-very-long-name-that-is-too-long", we shouldn't
match "this-is-a-very-" (but the old code would).
The cmdline code was also broken on Android because it used basename(3)
rather than getbasename. This doesn't affect glibc because there's a
workaround in portability.h to ensure that we get the non-POSIX basename(3)
with glibc but then a non-glibc section that ensures everyone else gets
POSIX basename(3). That should probably be removed (and maybe `basename`
poisoned) to prevent similar mistakes in future.
Bug: http://b/73123244
|
|
|
|
stuff syslog.h does into lib.c and portability.h
|
|
|
|
Found running LTP file system tests on Android.
Bug: http://b/70627145
|
|
Before:
toy: Unknown option p (See "toy --help")
After:
toy: Unknown option p (see "toy --help")
|
|
|
|
of utf8 support (not finished yet)...
Adds new loopfiles_lines() wrapper to lib.c that calls do_lines from loopfiles.
|
|
This avoids "xargs: exec echo: Argument list too long" errors in practice.
find(1) needs to be fixed too, but that's a bit more complicated and a working
xargs provides a workaround.
Bug: http://b/65818597
Test: find /proc | strace -f -e execve ./toybox xargs echo > /dev/null
|
|
program to compare against libc output.
|
|
scripts/*.c builds against lib.c but not linestack.c.
|
|
|
|
(According to the git history I added "b" for "od" but the man page says 512
there too.)
|
|
|
|
full help text, and put it back in the args constraint failures.
|
|
|
|
|
|
Recent-ish clang doesn't like self-assignment. Google/Android code always
uses the [template-based moral equivalent of] __attribute__((__unused__))
to keep both compilers happy.
|
|
Once upon a time you could call fchown() and let it fail. Then gcc decided
not using its return code was inconcievable, but you could typecast it to (void)
to shut it up. Then gcc noticed people doing that and clutched its pearls and
took it away, so I added an if() statement that does nothing with the result
because we _expect_ this to fail when we're not root. Then clang started
complaining about an if (); statement with the semicolon on the same line,
but decided it's ok if the ; is on the next line (I.E. significant whitespace
in C), so I'm adding an "assignemnt to self" that gets optimized away so
it does a more _explicit_ nothing (the same way you suppress gcc's broken
"this isn't used uninitialized" warnings).
If the compilers weren't going to so much trouble to force the issue I might
add code to only call fchown when we're UID 0, but I refuse to be coerced
into it. (And if getpid() is still a system call instead of a vdso member
then it doesn't actually _save_ us anything, the dentry should be hot and
the permission check was just "if (!uid)" before selinux entered into it and
we're operating on an fd so the security's the same.)
|
|
|
|
|
|
|
|
of number, but never NULL. Both returned in static buffer good through
next call.)
|
|
add xopenro() that takes one argument and understands "-" means stdin,
and switch over lots of users.
|
|
|
|
|
|
|
|
to use this (and tail) as examples of simplicity of implementation winning
and losing.
|
|
|
|
|
|
|
|
|
|
both did it) so use getbasename instead.
|
|
(The sigatexit() handler list code now calls the handlers on the normal
exit path, so _exit(1) from the handler is wrong; exit_signal() modifies
toys.exitval if necessary.)
|
|
|
|
major/minor/makedev, but glibc has vowed to break existing programs
(https://sourceware.org/ml/libc-alpha/2015-11/msg00253.html)
and replace it with _another_ non-standard header (not in posix or lsb),
so let's just add functions to lib/ that do the transform ourselves.
|
|
Also use finit_module if available. Given that "insmod -" requires
init_module, maybe this isn't a worthwhile optimization. Given that
"insmod /actual/file.ko" is the common use case, maybe it is.
Fix a bug in readfileat where *plen would be corrupted if you didn't supply
your own buffer (because ibuf is 0 in that case, not a pointer to the start
of the allocated space).
|
|
Unlike the POSIX file(1), there's no magic file here, just hard-coded
common (non-obsolete) file formats. Personally most of my use of file(1)
is as a one-line readelf(1) summarizer, so although I assume a full POSIX
file(1) is out of scope (because just the database would likely be larger
than all the rest of toybox), a subset that only supports in-use file types
actually covers most of the use cases I encounter personally.
Also fix peek_be/peek_le.
|
|
the deeply sad passwd heuristics that don't even check numbers and punctuation.
|
|
|
|
|
|
Also, I forgot to check in uuid_show() last time.
|
|
|
|
Some things are worse when you remove the 'p'.
|
|
Commit 4b4ab6a50998 broke loopfiles' handling of "-".
This broke the existing cat tests, so no new test is necessary here.
|
|
checking, and fix up format checking complaints.
Added out(type, value) function to stat to avoid a zillion printf typecasts.
|
|
|