| Age | Commit message (Collapse) | Author |
|---|
|
A TOYFLAG_NOFORK command must run in the context of toysh, but a MAYFORK
can either run standalone or run in the toysh process. MAYFORK means it
cleans up after itself: no leaked resources (malloc, mmap, filehandles, etc),
even in error_exit() paths that would longjmp() back to the shell. It also
doesn't discard anything we need to retain (don't close stdout, change
toys.optargs[] so we can't free it, etc)...
|
|
No current caller except grep needs this, but consistency seems like a
good idea.
Also change the xregcomp error message to be a bit more human-readable,
rather than mention an implementation detail.
|
|
|
|
Test: no EBADF in `strace -e close ./toybox tac /proc/version`
|
|
I started this last night, but thought I'd aim to send multiple small
patches rather than work through all the callers and send one big patch.
I've deliberately chosen the ugly name `allocated_length` because we've
had historical bugs where folks think this a line length in the sense of
the return value. I do wonder whether we should actually have some kind
of getline() wrapper that hides the `char *`/`size_t` pair in lib/,
which makes the function easier to use in most cases but does add the
less common gotcha that you wouldn't be able to getline() through
multiple files at once (which does happen in at least one toy).
But maybe the real fix is to look harder for places where we can just
use loopfiles_lines? Speaking of which, should we actually add two more
arguments to that? Specifically: switch it to getdelim() rather than
getline() behind the scenes, and also add a way to have the trailing
'\n' automatically removed, since that seems to be what most callers
want?
Anyway, that seemed like enough questions that it was time to send this
initial patch out before doing too much more...
|
|
Pointed out during code review of the recent refactor.
|
|
Before this patch, we're effectively doing `pidof -x` all the time. This
patch changes names_to_pid() to allow us to say whether or not we want to
include scripts, and adjusts the callers appropriately.
Also add tests for `pidof` versus `pidof -x` which pass after this
patch, without regressing the existing killall tests.
|
|
Rename the existing xbind/xconnect to xbindany/xconnectany, to make room
for new xbind/xconnect that are more like 'x' versions of the regular
bind and connect. Move explicit bind/connect callers over to
xbind/xconnect.
Of the affected commands, only netcat is actually used by Android. It
was the most recent patch to netcat that made the lack of a more
traditional xbind/xconnect apparent.
|
|
Investigating why the toybox tar tests fail on Android with toybox dd, I
realized I was rewriting a part of dd I'd rewritten before!
This is a re-send of my 2019-02-22 patch, rebased against the current
ToT...
This patch was originally motivated because after suggesting to the author of
https://stackoverflow.com/questions/17157820/access-vdsolinux/54797221#54797221
that he could tell dd to work in bytes rather than blocks, I realized
that our dd doesn't actually support that. But the rewrite of the main
loop is necessary to fix the incorrect output from the dd calls in the
tar test.
Without this patch, `yes | dd bs=65536 count=1 > fweep` basically gives
random output, based on how many bytes the pipe feels like giving you in
your first read. (As far as I know, dd *without* bs= was fine, but I
can't guarantee that that's true, just that I haven't seen it fail.)
Also switch to TAGGED_ARRAY and comma_* for conv rather than add two more
copies of an undesired idiom. It turned out -- contrary to the belief of
cp(1) -- that comma_scan isn't suitable for this because of its magic
handling of "no" prefixes. (It's actually harmless in cp because none
of the --preserve options begin with "no", but some dd options do.) To
this end, comma_remove is a less-magic comma_scan.
I've also changed an `if` to a `while` because other implementations
allow things like `--preserve=mode,mode` or `conv=sync,sync`. (If we
decide this is a bug rather than a feature, we should at least fix the
error message to be clear that we're rejecting the *duplication*, not
the option itself.)
I've also fixed the ^C behavior by simply adding a direct SIGINT handler
rather than trying to be clever inside the read loop (which is why we
weren't handling the SIGINT until the read returned).
I've also removed `strstarteq` and just added the '=' to each literal
when calling regular `strstart`.
Plus basic tests.
|
|
|
|
Change names_to_pid() so that we can actually match shell scripts with
long names (the code to get the shell script's name was correct, but
there was an extra test preventing us from actually comparing it to the
sought name).
In kill.c itself, remove a dead test for -l and switch to the FLAG()
macro.
Also extend the tests to explicitly cover long and short names.
|
|
This failure was a little hard to parse:
tar: Unknown option mode (see "tar --help")
This would have been clearer:
tar: Unknown option 'mode' (see "tar --help")
|
|
some other way.
|
|
|
|
Sets ->again |= 2 when that happens and ->st is zeroed.
While we're there, don't memset(st) and then memcpy(st) over it, and
use O_PATH in the open().
|
|
macOS doesn't have real-time signals, and it has a slightly different
set of non-portable signals from Linux.
With this, the toybox `kill -l` output matches the default macOS kill(1).
|
|
Include all the signals, not just the POSIX ones.
In particular, improve support for real-time signals. My attempt to
switch AOSP builds over to toybox timeout got reverted when I broke the
ART build bots which use SIGRTMIN+2.
Also fix `kill -l 3` to show "QUIT" and `kill -l QUIT` to show "3".
Also make the `kill -l` output include numbers and names, and format it
to look better on a 80x24 tty, because it's always August 1978 somewhere.
|
|
This patch adds a BSD version of xgetmountlist (for the path ==
NULL case only), tested on macOS. It also papers over the differences
between macOS' and Linux's xattr APIs. For once I think the macOS
one is better. The imitation of mknodat I've had to write swings
things back in Linux's favor though.
BSD calls f_frsize by the name f_iosize instead. (FWIW, it looks
like this is meaningless on Linux and actually meaningful on macOS.)
I've added one #if to toys/ --- I'm calling pathconf in stat.c to
work around the absence of f_namelen, and have left a TODO with an
explanation. I'm not sure what the best fix is here, so punting.
No-one can agree what f_fsid is, even if they're all basically the
same, so work around the `val` versus `__val` issue between macOS
and Linux.
With this patch, it's now possible to build cp/mv/install and stat
for macOS too. (Which completes the set of "toybox commands currently
used on Linux as part of the AOSP build" if you ignore stuff that
deals with processes, which I doubt we'll ever be able to support
for lack of any API.)
|
|
These are Android specific, so not really helping the outside
community, and are getting more and more Android dependencies to work
correctly, so let's drop these from toybox and build them within
Android.
Change-Id: Ic6022f1f506e10868c61f55d64fa4e7c1b14eba2
|
|
Trivial breakage from the recent refactoring.
Also add tail to the default config for macOS.
I'm still unconvinced that CONFIG_TAIL_SEEK makes sense in a world
where inotify support is always built in, but that's an argument
for another day...
|
|
|
|
This only implements the format specifiers that I've seen used in the
wild (which is actually a significant fraction of the total supported by
findutils' find). The most obvious gap is in the time support. I'm happy
to add more, but didn't want to add stuff "just because".
I'd say %A@, %C@, and -- for SELinux users -- %Z are probably the most
plausibly useful formats still missing. I don't think the human-readable
date formatting is particularly useful unless someone's seen it actually
used in the wild. The %T+ "full ISO" format being the most likely
exception to that.
Anyway, this is enough for me get started building AOSP with toybox find.
|
|
into a structure passed as arguments, add x prefix to functions that can fail,
add overflow test.
|
|
Factor out the inotify code and add a kqueue equivalent.
Specifically tested on macOS 10.14, but I assume this works for other
BSDs too, given that I worked from the FreeBSD man page...
|
|
|
|
|
|
|
|
|
|
|
|
This is a 15 year old freebsd extension (presumably thus also available on
MacOS) that glibc adopted in 2004, uClibc adopted in 2005, and bionic
supports. The only thing that DOESN'T support it is musl, once again
because its maintainer explicitly decided not to
(https://www.openwall.com/lists/musl/2013/01/15/26), so add an #ifdef
to let musl stay uniquely broken. (It'll stop at first NUL, everything
else can match NULs).
Finally fixes "s/x/y/g on a megabyte line of x's takes forever" issue.
|
|
Don't strlen() to find NUL to skip to until after we've confirmed first
section hasn't got a match (by calling regexec() on it).
|
|
|
|
Found when trying to update the toybox prebuilt used for the Android
build.
Also add the corresponding test.
|
|
They're forwarded to libprocessgroup, but we may as well go straight to
the source since neither library is in the NDK anyway.
This code is unfortunate because it means that even `toybox true` ends
up pulling in a JSON parser at runtime, because ps might call
get_sched_policy/get_sched_policy_name. I'll experiment with
dlopen-on-demand in portability.c and see whether the savings are
worthwhile, but for now at least use the current library directly so we
can save *one* dlopen!
|
|
libbuf instead of toybuf in lib/
|
|
|
|
|
|
leaking memory), and mod env command to test it.
|
|
|
|
|
|
Although we can get away with ignoring termcap/terminfo on the output
side by restricting ourselves to generally-supported escape sequences,
the input side is trickier because we need to support the sequences sent
by common terminals. Luckily, this isn't is as bad as it sounds because
only Home/End commonly differ. But it does mean we need a slightly
different implementation to deal with the many-to-one mapping.
Since we can't use TAGGED_ARRAY for this (without inflicting pain on all
the callers) I've also switched to OR-ing in the modifier keys, so we
have (say) KEY_UP|KEY_SHIFT rather than a separate KEY_SUP. This also
generalizes better should we ever need to support multiple modifiers at
once.
To reduce the number of #defines, I've also switched from KEY_F1,
KEY_F2, and so on to KEY_FN+1, KEY_FN+2, and so on. This isn't obviously
necessary, and easily undone if we'd rather have move #defines in return
for slightly more natural naming.
To enable all this, I've inverted scan_key and scan_key_getsize so that
scan_key_getsize is now the underlying function, and we don't waste all
the top bits encoding width and height between scan_key and
scan_key_getsize.
Tested by pressing Home and End in hexedit in all of the terminals
available to me.
|
|
|
|
and tweak a couple comments.
|
|
|
|
SELinux on Android is unhappy if you try to read "/":
avc: denied { read } for name="/" dev="dm-3" ino=2 scontext=u:r:hal_dumpstate_impl:s0 tcontext=u:object_r:rootfs:s0 tclass=dir permissive=0
That could happen via the open of ".." too, and potentially any other
directory might have similar restrictions, so move all of the open calls
to using O_PATH.
O_PATH seems more intention-revealing given what this function is doing anyway.
|
|
|
|
SELinux on Android is unhappy if you try to read "/":
avc: denied { read } for name="/" dev="dm-3" ino=2 scontext=u:r:hal_dumpstate_impl:s0 tcontext=u:object_r:rootfs:s0 tclass=dir permissive=0
O_PATH seems more intention-revealing anyway.
|
|
and fix tests to pass on host too.
|
|
|
|
Now there's a second user... the libc function already null terminates
at len+1, and it doesn't malloc the full size if strlen() smaller so
the redundant termination stomped unallocated memory. Oops.
sort.c never noticed because it calculated length to truncate or copy
existing string, so never hit this.
|
