]> git.lizzy.rs Git - plan9front.git/log
plan9front.git
10 years agopc kernel: split mpstartap() and squidboy into separate file... stuff for amd64
cinap_lenrek [Sat, 1 Feb 2014 09:23:17 +0000 (10:23 +0100)]
pc kernel: split mpstartap() and squidboy into separate file... stuff for amd64

10 years agokernel: usb fixes for amd64
cinap_lenrek [Sat, 1 Feb 2014 09:20:43 +0000 (10:20 +0100)]
kernel: usb fixes for amd64

10 years agokernel: fix pointer truncation in xspanalloc(), fix format prints
cinap_lenrek [Sat, 1 Feb 2014 09:17:53 +0000 (10:17 +0100)]
kernel: fix pointer truncation in xspanalloc(), fix format prints

10 years agokernel: handle amd64 40 byte headers in exec()
cinap_lenrek [Sat, 1 Feb 2014 09:16:55 +0000 (10:16 +0100)]
kernel: handle amd64 40 byte headers in exec()

10 years agovesa: explicitely use /386/include/ureg.h
cinap_lenrek [Sat, 1 Feb 2014 09:14:41 +0000 (10:14 +0100)]
vesa: explicitely use /386/include/ureg.h

10 years agolibmach: pc64 uses 4K pages, IP -> PC
cinap_lenrek [Sat, 1 Feb 2014 08:55:34 +0000 (09:55 +0100)]
libmach: pc64 uses 4K pages, IP -> PC

10 years agolibmemdraw: apply erik quanstros fix for sign preserving in byteaddr()
cinap_lenrek [Sat, 1 Feb 2014 08:54:09 +0000 (09:54 +0100)]
libmemdraw: apply erik quanstros fix for sign preserving in byteaddr()

10 years agolibdraw: use pid as initial screenid so one can have more than 25 windows.
cinap_lenrek [Sat, 1 Feb 2014 08:53:08 +0000 (09:53 +0100)]
libdraw: use pid as initial screenid so one can have more than 25 windows.

10 years ago6l: fix warning, setmalloctag declaration, missing header type cases
cinap_lenrek [Sat, 1 Feb 2014 08:52:06 +0000 (09:52 +0100)]
6l: fix warning, setmalloctag declaration, missing header type cases

10 years ago/lib/face: add some new faces
stanley lieber [Wed, 22 Jan 2014 06:20:40 +0000 (01:20 -0500)]
/lib/face: add some new faces

10 years agofortunes: undelete
stanley lieber [Wed, 22 Jan 2014 06:19:50 +0000 (01:19 -0500)]
fortunes: undelete

10 years agosdiahci: fix hardcoded 512 byte sector size
cinap_lenrek [Wed, 22 Jan 2014 19:12:19 +0000 (20:12 +0100)]
sdiahci: fix hardcoded 512 byte sector size

10 years agokernel: fix ulong abuse in xalloc
cinap_lenrek [Tue, 21 Jan 2014 21:12:25 +0000 (22:12 +0100)]
kernel: fix ulong abuse in xalloc

10 years agokernel: convert putmmu() to uintptr for va and pa
cinap_lenrek [Mon, 20 Jan 2014 02:17:55 +0000 (03:17 +0100)]
kernel: convert putmmu() to uintptr for va and pa

10 years agokernel: various cleanups
cinap_lenrek [Mon, 20 Jan 2014 01:16:42 +0000 (02:16 +0100)]
kernel: various cleanups

10 years agomalloc(2): update alloctag types
cinap_lenrek [Sun, 19 Jan 2014 23:54:36 +0000 (00:54 +0100)]
malloc(2): update alloctag types

10 years agokernel: apply uintptr for ulong when a pointer is stored
cinap_lenrek [Sun, 19 Jan 2014 23:47:55 +0000 (00:47 +0100)]
kernel: apply uintptr for ulong when a pointer is stored

this change is in preparation for amd64. the systab calling
convention was also changed to return uintptr (as segattach
returns a pointer) and the arguments are now passed as
va_list which handles amd64 arguments properly (all arguments
are passed in 64bit quantities on the stack, tho the upper
part will not be initialized when the element is smaller
than 8 bytes).

this is partial. xalloc needs to be converted in the future.

10 years agomalloc: change malloc and realloc tag types to uintptr
cinap_lenrek [Sun, 19 Jan 2014 23:09:53 +0000 (00:09 +0100)]
malloc: change malloc and realloc tag types to uintptr

10 years agoape: remove local copy of memccpy()
cinap_lenrek [Sun, 19 Jan 2014 23:05:19 +0000 (00:05 +0100)]
ape: remove local copy of memccpy()

10 years agokernel: fix declaration of getcallerpc()
cinap_lenrek [Sun, 19 Jan 2014 02:32:50 +0000 (03:32 +0100)]
kernel: fix declaration of getcallerpc()

10 years agokernel: replace BY2WD in scanline width calculation to sizeof(ulong) like libmemdraw
cinap_lenrek [Sun, 19 Jan 2014 02:29:52 +0000 (03:29 +0100)]
kernel: replace BY2WD in scanline width calculation to sizeof(ulong) like libmemdraw

10 years agolibmach: fix wrong operand order of MOVLQSX instruction disassembly
cinap_lenrek [Fri, 17 Jan 2014 20:09:17 +0000 (21:09 +0100)]
libmach: fix wrong operand order of MOVLQSX instruction disassembly

10 years agoCorrected some man pages.
aap [Fri, 17 Jan 2014 18:15:24 +0000 (19:15 +0100)]
Corrected some man pages.

10 years agodevdraw: fix memory corruption reading draw ctl file
cinap_lenrek [Mon, 13 Jan 2014 23:22:13 +0000 (00:22 +0100)]
devdraw: fix memory corruption reading draw ctl file

when user does read of exactly 12*12 bytes on draw
ctl file, the snprint() adds one more \0 byte writing
beyond the user buffer and corrupting memory.

fix this by not snprint()ing the final space and add
it manually.

10 years agomerge
mischief [Tue, 14 Jan 2014 02:06:58 +0000 (18:06 -0800)]
merge

10 years agohpost: fix Content-Type parameter separator for multipart/form-data POSTs
mischief [Tue, 14 Jan 2014 01:57:45 +0000 (17:57 -0800)]
hpost: fix Content-Type parameter separator for multipart/form-data POSTs

10 years agoadd /lib/uriel; delete fortunes
stanley lieber [Sun, 12 Jan 2014 18:00:02 +0000 (13:00 -0500)]
add /lib/uriel; delete fortunes

10 years agoadded a segment for segattach(2) that gives access to the raspberry pi's gpio header.
Matthew Veety [Sun, 12 Jan 2014 16:00:38 +0000 (11:00 -0500)]
added a segment for segattach(2) that gives access to the raspberry pi's gpio header.

10 years agoetheryuk: fix lockups (thanks burnzez for testing)
cinap_lenrek [Sun, 12 Jan 2014 11:08:10 +0000 (12:08 +0100)]
etheryuk: fix lockups (thanks burnzez for testing)

according to a comment in linux driver, reading Isrc2
register caused interrupts to be disabled. we used
to read Isrc2 in ifstat() and it was confirmed that
reading ifstat locks up ethernet. removing the Isrc2
read in ifstats, and also reenable interrupts after
reading Isrc2 when the interrupt was not for us.
(this is from the linux driver)

in replenish(), set ring software write pointer (Sring.wp)
*before* the hardware write index register. otherwise
rx() could get status notification for completed
receive but wont find the rx descriptor in the ring.

handle uint wrap arround when calculating ring fill
count and remaining count.

10 years agonetif: fix wrong qid in 3rd level stats/ifstats files (thanks burnzez)
cinap_lenrek [Sat, 11 Jan 2014 15:07:35 +0000 (16:07 +0100)]
netif: fix wrong qid in 3rd level stats/ifstats files (thanks burnzez)

the stats and ifstats files in the 3rd level of a netif
are not per connection, but for the interface.

this made fstat fail for /net/ether0/N/*stats where N > 0
as the NETID() bits in the qid didnt compare.

10 years agorio: properly handle initial wrap arround in wlook()
cinap_lenrek [Sat, 11 Jan 2014 09:46:07 +0000 (10:46 +0100)]
rio: properly handle initial wrap arround in wlook()

10 years agorio: implement "look" (thanks lf94)
cinap_lenrek [Sat, 11 Jan 2014 09:14:35 +0000 (10:14 +0100)]
rio: implement "look" (thanks lf94)

10 years agokernel: fix fairshare formula in comment (thanks erik)
cinap_lenrek [Fri, 10 Jan 2014 14:12:28 +0000 (15:12 +0100)]
kernel: fix fairshare formula in comment (thanks erik)

10 years agomixfs: make mixbuffer addition atomic
cinap_lenrek [Fri, 10 Jan 2014 13:51:29 +0000 (14:51 +0100)]
mixfs: make mixbuffer addition atomic

serialize mixing with spinlock to make the addition
on the mixbuffer samples atomic.

10 years agomerge
cinap_lenrek [Fri, 10 Jan 2014 00:20:24 +0000 (01:20 +0100)]
merge

10 years agolibdraw: cleanup string() and stringwidth()
cinap_lenrek [Fri, 10 Jan 2014 00:19:14 +0000 (01:19 +0100)]
libdraw: cleanup string() and stringwidth()

getting rid of the goto at the end of the while
loop by moving the if(subfontnae) case before
the cachechars() call.

10 years agowinwatch: this is not about sanity. rows might be negative
mischief [Thu, 9 Jan 2014 05:32:27 +0000 (21:32 -0800)]
winwatch: this is not about sanity. rows might be negative

10 years agowinwatch: prevent divide by zero if the window height is too small
mischief [Thu, 9 Jan 2014 04:52:40 +0000 (20:52 -0800)]
winwatch: prevent divide by zero if the window height is too small

10 years agomothra: remove version
cinap_lenrek [Tue, 7 Jan 2014 01:57:39 +0000 (02:57 +0100)]
mothra: remove version

10 years agolibframe: fix _frcanfit() for zero width runes, simplify chopframe()
cinap_lenrek [Tue, 7 Jan 2014 01:24:16 +0000 (02:24 +0100)]
libframe: fix _frcanfit() for zero width runes, simplify chopframe()

10 years agoweather: trim garbage from output
stanley lieber [Mon, 6 Jan 2014 19:55:42 +0000 (14:55 -0500)]
weather: trim garbage from output

10 years agolibframe: avoid _frcanfit() call in frdelete()
cinap_lenrek [Mon, 6 Jan 2014 02:56:45 +0000 (03:56 +0100)]
libframe: avoid _frcanfit() call in frdelete()

no need to calculate number of runes to fit in
box when box contains no runes.

10 years agolibdraw: fix stringwidth problems
cinap_lenrek [Mon, 6 Jan 2014 02:49:14 +0000 (03:49 +0100)]
libdraw: fix stringwidth problems

cachechars() used to skip over characters on its own when
loadchar() could not find the character or a PJW replacement.
this resulted in wrong width calculation. now we just return
and handle the case inside _string and _stringwidth.

fix subfont leak in stringwidth()

remove annoying prints in stringwidth()

10 years agomerge
cinap_lenrek [Sun, 5 Jan 2014 01:39:23 +0000 (02:39 +0100)]
merge

10 years agorio: make scrollwheel handling consistent with sam
cinap_lenrek [Sun, 5 Jan 2014 01:37:42 +0000 (02:37 +0100)]
rio: make scrollwheel handling consistent with sam

scrollwheel now works proportional to y mouse position.
special case is when shift is hold down. then scrollwheel
will work like before and scroll one line up or down.

10 years agofortunes: This is Go.
stanley lieber [Sat, 4 Jan 2014 18:13:38 +0000 (13:13 -0500)]
fortunes: This is Go.

10 years ago9fs: add atom and atomdump (fix the mount spec)
cinap_lenrek [Fri, 3 Jan 2014 21:53:21 +0000 (22:53 +0100)]
9fs: add atom and atomdump (fix the mount spec)

10 years ago9fs: add atom alias for atom.9atom.org
cinap_lenrek [Fri, 3 Jan 2014 21:25:55 +0000 (22:25 +0100)]
9fs: add atom alias for atom.9atom.org

10 years agokernel: simplify procalarm()
cinap_lenrek [Fri, 3 Jan 2014 01:16:18 +0000 (02:16 +0100)]
kernel: simplify procalarm()

10 years agoalarm: skip timed out alarms when inserting in procalarm() (thanks erik)
cinap_lenrek [Fri, 3 Jan 2014 00:40:17 +0000 (01:40 +0100)]
alarm: skip timed out alarms when inserting in procalarm() (thanks erik)

from erik quanstros 9fans post:

i think the list insertion code needs a single-read
test that f->alarm != 0. to prevent the 0 from
acting like a fencepost.  e.g. trying to insert -10 into
list -40 -30 0 -20.

if(alarms.head) {
l = &alarms.head;
for(f = *l; f; f = f->palarm) {
>> fw = f->alarm;
>> if(fw != 0 && (long)(fw - when) >= 0) {
up->palarm = f;
*l = up;
goto done;
}
l = &f->palarm;
}
*l = up;
}

10 years agops2mouse: retry disable packet streaming command 0xf5
cinap_lenrek [Thu, 2 Jan 2014 10:07:06 +0000 (11:07 +0100)]
ps2mouse: retry disable packet streaming command 0xf5

disabling mouse packet streaming command 0xf5 can fail
when a packet is currently transmitted.

this can be seen when one moves the mouse while running:

while(){echo accelerated >/dev/mousectl; sleep 0.5}

10 years agops2mouse: disable packet streaming during reprogramming
cinap_lenrek [Thu, 2 Jan 2014 06:44:17 +0000 (07:44 +0100)]
ps2mouse: disable packet streaming during reprogramming

sometimes, ps2 mouse setup failed on thinkpads during boot. disabling
packet streaming while we program the mouse seems to fix the issue.

10 years agomerge
cinap_lenrek [Wed, 1 Jan 2014 06:41:06 +0000 (07:41 +0100)]
merge

10 years agokernel: nil check, exited procs handling in postnote()
cinap_lenrek [Wed, 1 Jan 2014 06:39:17 +0000 (07:39 +0100)]
kernel: nil check, exited procs handling in postnote()

make sure not to dereference Proc* nil pointer. this can potentially
happen from devip which has code like:

if(er->read4p)
postnote(er->read4p, 1, "unbind", 0);

the process it is about to kill can zero er->read4p at any time,
so there is the possibility of the condition to be true and then
er->read4p becoming nil.

check if the process has already exited (p->pid == 0) in postnote()
under p->debug qlock.

10 years agoproc(3): the args file is writable, not read only (thanks qrstuv)
cinap_lenrek [Wed, 1 Jan 2014 05:44:48 +0000 (06:44 +0100)]
proc(3): the args file is writable, not read only (thanks qrstuv)

10 years agokernel: remove error label in pgrpnote() arround postnote
cinap_lenrek [Wed, 1 Jan 2014 04:16:41 +0000 (05:16 +0100)]
kernel: remove error label in pgrpnote() arround postnote

postnote does not raise error, so the error label is not
neccesary.

10 years agokernel: fix alarm postnote race
cinap_lenrek [Wed, 1 Jan 2014 04:14:38 +0000 (05:14 +0100)]
kernel: fix alarm postnote race

when alarmkproc is commited to send the alarm note to the process,
the process might have exited already, or worse, being reused for
another process. pexit() zeros p->alarm at the beginning, but the
kalarmproc() might read p->alarm before pexit() zeroed it, decide
to send the note, then get preempted and pexit() releases the proc.
once kalarmproc() is resumed, the proc might be already something
different and we send the note to the wrong thing.

we now check p->alarm under the debug qlock. that way, pexit()
cannot make progress while we test the condition.

remove the error label arround postnote(). postnote does not raise
error.

10 years agodevproc: fix noteid permission checks for none
cinap_lenrek [Tue, 31 Dec 2013 11:23:55 +0000 (12:23 +0100)]
devproc: fix noteid permission checks for none

make sure noteid is valid (>0).

prohibit changing note group of kernel processes. this is also
checked for in pgrpnote().

prevent "none" user from changing its note group to another "none"
sessions. this would allow him to send notes other none processes
other than its own.

10 years agokernel: check for error label stack underflow in nexterror()
cinap_lenrek [Tue, 31 Dec 2013 06:29:55 +0000 (07:29 +0100)]
kernel: check for error label stack underflow in nexterror()

10 years agomothra: mouse scrollwheel behave the same as in sam
stanley lieber [Tue, 31 Dec 2013 05:52:52 +0000 (00:52 -0500)]
mothra: mouse scrollwheel behave the same as in sam

10 years agokernel: halt idle processors on mp system by default (from sources)
cinap_lenrek [Tue, 31 Dec 2013 03:41:51 +0000 (04:41 +0100)]
kernel: halt idle processors on mp system by default (from sources)

one can add:

int idle_spin = 1;

in the kernel configuration to enable the old behaviour. see the
comment at idlehands().

10 years agokernel: remove unused debug cprint() function from port/cache.c
cinap_lenrek [Sun, 29 Dec 2013 06:50:58 +0000 (07:50 +0100)]
kernel: remove unused debug cprint() function from port/cache.c

the cache can be inspected more easily with acid kernel library.

10 years agokernel: dont call pprint() while holding up->debug qlock
cinap_lenrek [Sun, 29 Dec 2013 06:48:19 +0000 (07:48 +0100)]
kernel: dont call pprint() while holding up->debug qlock

pprint() might block or even (maliciously) call into
devproc write which will corrupt the qlock chain on attempt
to qlock up->debug again.

10 years agokernel: make sure user text, data and bss wont overlap the stack segment in sysexec()
cinap_lenrek [Sun, 29 Dec 2013 05:11:18 +0000 (06:11 +0100)]
kernel: make sure user text, data and bss wont overlap the stack segment in sysexec()

10 years agomothra: add regular expression text search (thanks mischief, sl)
cinap_lenrek [Sat, 28 Dec 2013 20:48:53 +0000 (21:48 +0100)]
mothra: add regular expression text search (thanks mischief, sl)

this works differently from mischiefs original patch. instead of
overloading the address bar, we popup our own enter box. the
function can be invoked from the menu or by hitting ^F.

10 years agopage: fix page bookmarks for epub files
cinap_lenrek [Sat, 28 Dec 2013 05:49:51 +0000 (06:49 +0100)]
page: fix page bookmarks for epub files

10 years agoape: move strdup() from libbsd to libap (from sources)
cinap_lenrek [Sat, 28 Dec 2013 04:14:10 +0000 (05:14 +0100)]
ape: move strdup() from libbsd to libap (from sources)

including <string.h> should be enougth to make strdup()
available.

10 years agocifs(4): replace with correct version
stanley lieber [Fri, 27 Dec 2013 21:42:36 +0000 (16:42 -0500)]
cifs(4): replace with correct version

10 years agocifs(4), cifsd(8): add HISTORY
stanley lieber [Fri, 27 Dec 2013 21:32:37 +0000 (16:32 -0500)]
cifs(4), cifsd(8): add HISTORY

10 years agoadd HISTORY to various man pages
stanley lieber [Fri, 27 Dec 2013 21:22:05 +0000 (16:22 -0500)]
add HISTORY to various man pages

10 years agoaudio/mixfs: implement Tstat to obtain buffer size, better underrun handling
cinap_lenrek [Fri, 27 Dec 2013 19:39:56 +0000 (20:39 +0100)]
audio/mixfs: implement Tstat to obtain buffer size, better underrun handling

10 years agoaudio(1): add pcmconv to HISTORY
stanley lieber [Thu, 26 Dec 2013 21:22:23 +0000 (16:22 -0500)]
audio(1): add pcmconv to HISTORY

10 years agofortunes: That's a video card, Blake.
stanley lieber [Thu, 26 Dec 2013 21:11:05 +0000 (16:11 -0500)]
fortunes: That's a video card, Blake.

10 years agoaudio(3): fix spelling
stanley lieber [Thu, 26 Dec 2013 21:07:01 +0000 (16:07 -0500)]
audio(3): fix spelling

10 years agoaudio(1): add HISTORY, fix spelling
stanley lieber [Thu, 26 Dec 2013 21:06:37 +0000 (16:06 -0500)]
audio(1): add HISTORY, fix spelling

10 years agoadd audio/mixfs to allow simultanious playback streams over a single /dev/audio
cinap_lenrek [Thu, 26 Dec 2013 20:04:25 +0000 (21:04 +0100)]
add audio/mixfs to allow simultanious playback streams over a single /dev/audio

10 years agojson(2): fix out of order SOURCE heading
cinap_lenrek [Wed, 25 Dec 2013 19:09:00 +0000 (20:09 +0100)]
json(2): fix out of order SOURCE heading

10 years agodevmouse: exclude sroll-wheel buttons from differential button handling
cinap_lenrek [Wed, 25 Dec 2013 15:52:55 +0000 (16:52 +0100)]
devmouse: exclude sroll-wheel buttons from differential button handling

preventing wheel buttons 4 and 5 fron getting stuck
with multiple input devices.

10 years agoacid leak: remove arena pointer a < 0xff000000 check
cinap_lenrek [Wed, 25 Dec 2013 14:00:39 +0000 (15:00 +0100)]
acid leak: remove arena pointer a < 0xff000000 check

in the kernel, area can lie above 0xff000000

10 years agodevdraw: fix drawdelname() string memory leak
cinap_lenrek [Wed, 25 Dec 2013 13:55:16 +0000 (14:55 +0100)]
devdraw: fix drawdelname() string memory leak

10 years agowindow: fix various problems, cleanup
cinap_lenrek [Wed, 25 Dec 2013 00:30:43 +0000 (01:30 +0100)]
window: fix various problems, cleanup

with window -m, properly remove the old wsys from /dev before
binding new window to it, so the original window wont leak
into the new namespace.

do not pass -pid when $wsys does not refer to a local running rio.

cleanup environment before running sub-process.

set window label consistentently no matter if -m is used or not.

10 years agopage: scrollwheel, better backwards scrolling, fix blit bug when page was translated...
cinap_lenrek [Mon, 23 Dec 2013 14:38:52 +0000 (15:38 +0100)]
page: scrollwheel, better backwards scrolling, fix blit bug when page was translated offscreen

10 years ago/sys/lib/plumb/basic: add commas to some rules for file names that will frequently...
stanley lieber [Sun, 22 Dec 2013 19:56:04 +0000 (14:56 -0500)]
/sys/lib/plumb/basic: add commas to some rules for file names that will frequently contain commas

10 years agofshalt: rfork n to avoid limbo namespace when the command fails
cinap_lenrek [Sun, 22 Dec 2013 16:26:31 +0000 (17:26 +0100)]
fshalt: rfork n to avoid limbo namespace when the command fails

10 years agomerge
cinap_lenrek [Sat, 21 Dec 2013 17:06:24 +0000 (18:06 +0100)]
merge

10 years agoadd plumbing rules for page bookmarks and audio playback
cinap_lenrek [Sat, 21 Dec 2013 17:04:50 +0000 (18:04 +0100)]
add plumbing rules for page bookmarks and audio playback

10 years agopage: implement bookmarks (thanks trav_ for starting it)
cinap_lenrek [Sat, 21 Dec 2013 17:04:14 +0000 (18:04 +0100)]
page: implement bookmarks (thanks trav_ for starting it)

add new functions pageaddr() that returns a string describing
the page to be loaded. it is in the form of:

/path/to/file!pagename!subpage!....

one can jump to such a page by calling trywalk(name, addr)
where name and addr get concatinated with ! to form a page
address and then the currently loaded pages are walked up
to the nearest page which is then returned. (or nil when
not found). the remaining address will be set in the global
pagewalk variable.

once pages get loaded (asynchronously), pagewalk1() gets
called again on addpage() and continues the walking up to the
last page.

new program flag -j <addr> was added to jump to a page on
startup.

page address (without filename) can also be supplied in
plumb message with the "addr" attribute.

10 years agofortunes: What an amazing year for the Web.
stanley lieber [Thu, 19 Dec 2013 20:08:28 +0000 (15:08 -0500)]
fortunes: What an amazing year for the Web.

10 years agojson(2): fix spelling error
stanley lieber [Thu, 19 Dec 2013 19:59:50 +0000 (14:59 -0500)]
json(2): fix spelling error

10 years agoadd medium to low quality json(2) manual page
cinap_lenrek [Thu, 19 Dec 2013 18:56:04 +0000 (19:56 +0100)]
add medium to low quality json(2) manual page

10 years agorio: dont consider shell characters as part of the filename for auto completion
cinap_lenrek [Thu, 19 Dec 2013 17:12:16 +0000 (18:12 +0100)]
rio: dont consider shell characters as part of the filename for auto completion

rio looks backwards in the line for the beginning of a filename
that needs to be completed with ^F. this change makes the
characters: =, ^, ( and { stoppers, so filename completion
will work in all these cases:

foobar=/foo/ba^F; for(i in (fo^F ba^F)){/bin/baz^F

this means completion will not work for prefixes having these
special characters in them.

thanks to burnzez for bringing it up.

10 years agoacid leak: handle ALIGN_MAGIC padding of skip area for mallocalign()
cinap_lenrek [Tue, 17 Dec 2013 14:40:12 +0000 (15:40 +0100)]
acid leak: handle ALIGN_MAGIC padding of skip area for mallocalign()

the alloc and realloc tag words do not start at the beginning of
a block created by mallocalign(). instead, the block is padded
with ALIGN_MAGIC.

10 years agopool: update debug log format string for poolallocalign()
cinap_lenrek [Tue, 17 Dec 2013 14:32:48 +0000 (15:32 +0100)]
pool: update debug log format string for poolallocalign()

10 years agodevether: fix spelling mistype
cinap_lenrek [Tue, 17 Dec 2013 11:48:17 +0000 (12:48 +0100)]
devether: fix spelling mistype

10 years agoether82598, ether82563: remove packblock() avoidance hack
cinap_lenrek [Mon, 16 Dec 2013 20:47:06 +0000 (21:47 +0100)]
ether82598, ether82563: remove packblock() avoidance hack

10 years agoetheryuk: fix problems with yukon2 ep+ rev0, deoptimize
cinap_lenrek [Mon, 16 Dec 2013 20:39:22 +0000 (21:39 +0100)]
etheryuk: fix problems with yukon2 ep+ rev0, deoptimize

10 years agodevether: remove qfull prints and fix loopback packet handling of etheroq()
cinap_lenrek [Mon, 16 Dec 2013 18:19:15 +0000 (19:19 +0100)]
devether: remove qfull prints and fix loopback packet handling of etheroq()

dont spam the console with qfull warnings. this makes things worse.

handle loopback packets as stated in the comment. we call etheriq()
with fromwire=1 for loopback packets so etheriq() can pass the packet
on (without copying) or free it. dont inhibit interrupts while calling
etheriq(). etheriq() can safely be called from process and interrupt
context. it is unclear what this was supposed to fix and testing didnt
seem to have any odd effects.

10 years agodevssl: use parsecmd() to process control message to get rid of 128 byte stack buffer...
cinap_lenrek [Mon, 16 Dec 2013 09:06:34 +0000 (10:06 +0100)]
devssl: use parsecmd() to process control message to get rid of 128 byte stack buffer limit

 (11:02:29 PM) me: why is buf in /sys/src/9/port/devssl.c:/^sslwrite only 128 bytes?
 (11:02:58 PM) me: it makes it so you can't use a 128 bytes secret as negotiated by infauth in a secretin or secretout ctl message
 (11:03:30 PM) me: which in turn means you can't use such a secret with pushssl(2)
 (11:06:15 PM) me: inferno's sslwrite is limited to 32 bytes, but its ssl library writes to the secret files instead of to the ctl file
 (11:08:50 PM) mischief: what should it be instead of 128 bytes
 (11:08:58 PM) me: larger
 (11:09:16 PM) mischief: how about 129 bytes?
 (11:09:59 PM) me: also broken in 9front, by the way
 (11:15:14 PM) me: i guess it should be replaced with parsecmd

10 years agoexportfs: get rid of limits, cleanup
cinap_lenrek [Sun, 15 Dec 2013 06:49:53 +0000 (07:49 +0100)]
exportfs: get rid of limits, cleanup

get rid of the service buffer limit. keep service buffers
on a global freelist protected by lock.

dont fatal when we hit the process limit. instead, just
abort the rpc with an error.

handle rendezvous() interrupts.

10 years agogod damn it forgot fucking fns.h from my last commit
Matthew Veety [Fri, 13 Dec 2013 07:24:15 +0000 (02:24 -0500)]
god damn it forgot fucking fns.h from my last commit