this applet needlessly big! TODO: bb_getopt_ulflags()- use atol for the offset, which is always in bytes. Small step in eventually dropping bb_xparse_number().
- shrink simple obscure stuff a tiny bit: text data bss dec hex filename
789 0 0 789 315 obscure.o.oorig
771 0 0 771 303 obscure.o
- replace bzero by memset while at it.
- remove warning about redeclaration of __constant_htons- remove warning about format '%ld' expects type 'long int *', but argument #num
has type 'long unsigned int *'
Frank Sorenson added hotplug support to mdev. (I tweaked it a bit. Needto come up with a test suite for all the stuff that requires root access.
Something involving User Mode Linux or QEMU, probably...)
Ok, ping needs a rewrite from the ground up. (This is not a busybox app.)Fix bug 309, where ping -s 1 localhost would give an elapsed time 9 digits
long due to a stack overflow. Apparently, iputils also does this.
With -a, the jump to mount_it_now can skip the initialization of f, and if we don't zero it after closing it we re-close a filehandle that isn't open, and since this is a file _pointer_ it segfaults on a double free.Yeah, subtle bug. I need to break this out into separate functions if I can
figure out how to avoid making the code larger while doing so. Part of
the general -a and -o remount work I need to do, but that's after 1.1.0...
Bug 112, return value of awk should be /256. In reality, we should probablybe using WEXITSTATUS(), but until I can figure out why the heck that would
want to do (*(int *) &(status)) on the value, I'm happy just fixing the bug we
actually see.
Bug 547: writing out the tar file header before we confirm we can actuallyopen and read from the file isn't something we can recover from after the
fact. Resequence things to check first, write second.
Tito writes: Somehow while applying the bb_do_delay patch a change slipped in libbb.h that broke compilation.libbb.h Line 355
extern char bb_path_mtab_file[];
This conflicts with mtab_file.c
#if defined(CONFIG_FEATURE_MTAB_SUPPORT)
const char bb_path_mtab_file[] = "/etc/mtab";
#else
const char bb_path_mtab_file[] = "/proc/mounts";
#endif
Bug 624 wants quoted char support for printf, so you can do something like: printf '%d\n' '"x"'
and have it print out 120. This is the smallest implementation I can think
of at the moment.
No, we _want_ the suid and sgid bits reverted by chown, for security reasons.They mean something different when the user they're switching to is different,
so if you still want suid you reset it afterwards. This is a safety feature.
Jason Schoon writes:Here is a patch for kill that I posted long ago, but forgot about
until today. This allows kill to specify a negative process/group
(such as -1). The shell already had this fix applied by Vodz some
time ago.
Second attempt to simplify the login menu. Shadow password stuff at the top,with no dependencies (since depending on 9 of the 13 apps in the menu gets
a bit silly). And fix my attempt to make internal shadow password support
automatically depend on internal password support plus shadow password
support.
Some minor cleanups/bugfixes split off from the big remount work: Shorten GPL boilerplate.
Enabling FEATURE_CLEAN_UP broke things in two places.
Move the NFS with uClibc check to nfsmount.c
Shadow password support went beyond the dependency event horizon, just makeit an independent menu. And make internal shadow password support a subset
of the other internal password function support.
"Vladimir N. Oleynik"
Rob Landley
Paul Fox
Bernhard Reutner-Fischer
Mike Frysinger