]> git.lizzy.rs Git - plan9front.git/log
plan9front.git
11 years agodevsd: initialize unit->sense[0] in sdsetsense() (from erik quanstroms 9atom)
cinap_lenrek [Thu, 2 May 2013 21:12:37 +0000 (23:12 +0200)]
devsd: initialize unit->sense[0] in sdsetsense() (from erik quanstroms 9atom)

11 years agochan: fix potential path leak on clone in namec() (from erik quanstroms 9atom)
cinap_lenrek [Wed, 1 May 2013 19:56:02 +0000 (21:56 +0200)]
chan: fix potential path leak on clone in namec() (from erik quanstroms 9atom)

11 years agodevpipe: fix channel and queue leaks (from erik quanstroms 9atom)
cinap_lenrek [Wed, 1 May 2013 17:03:56 +0000 (19:03 +0200)]
devpipe: fix channel and queue leaks (from erik quanstroms 9atom)

11 years agoaoe: updating aoe driver from erik quanstros 9atom
cinap_lenrek [Wed, 1 May 2013 16:50:07 +0000 (18:50 +0200)]
aoe: updating aoe driver from erik quanstros 9atom

  1  the config string was grabbed Aoehsz too far into the packet due to using the wrong pointer to start.
  2  never accept a response with tag Tmgmt or Tfree.
  3  defend against "malicious" responses; ones with a response Aoehdr.type != request Aoehdr.type. this previously could
 cause the initiator to crash.
  4  vendor commands were improperly filtered out.

11 years agodevloopback: fix potential channel leak on allocation error (from eriks 9atom)
cinap_lenrek [Wed, 1 May 2013 16:31:13 +0000 (18:31 +0200)]
devloopback: fix potential channel leak on allocation error (from eriks 9atom)

11 years agocc: accept 24 bit numeric runes
cinap_lenrek [Wed, 1 May 2013 14:55:11 +0000 (16:55 +0200)]
cc: accept 24 bit numeric runes

11 years agofrexp: handle NaN values (from sources)
cinap_lenrek [Wed, 1 May 2013 14:44:04 +0000 (16:44 +0200)]
frexp: handle NaN values (from sources)

11 years agouartpci: add detection for StarTech PCI8S9503V (P588UG) (from sources)
cinap_lenrek [Wed, 1 May 2013 14:29:58 +0000 (16:29 +0200)]
uartpci: add detection for StarTech PCI8S9503V (P588UG) (from sources)

11 years agorio: fix highlight text for -b
stanley lieber [Wed, 1 May 2013 01:03:12 +0000 (21:03 -0400)]
rio: fix highlight text for -b

11 years agoblogcom +x
aiju [Tue, 30 Apr 2013 15:02:14 +0000 (17:02 +0200)]
blogcom +x

11 years agoadded blogcom
aiju [Tue, 30 Apr 2013 14:57:27 +0000 (16:57 +0200)]
added blogcom

11 years ago5c: dont emit conditional integer division instructions
cinap_lenrek [Mon, 29 Apr 2013 18:15:40 +0000 (20:15 +0200)]
5c: dont emit conditional integer division instructions

the integer division instructions are emulated with _div()
function patched in by 5l which does not handle conditional
execution. so do not optimize away the branch in that case.

11 years agoFix double-free fault when isaconfig returns nothing for "etherN". (from richard...
cinap_lenrek [Mon, 29 Apr 2013 16:39:44 +0000 (18:39 +0200)]
Fix double-free fault when isaconfig returns nothing for "etherN". (from richard millers devether-doublefree patch)

11 years ago/sys/lib/rootstub: fix missing /alpha/lib/ape directory
ppatience0 [Mon, 29 Apr 2013 15:12:47 +0000 (11:12 -0400)]
/sys/lib/rootstub: fix missing /alpha/lib/ape directory

11 years agobootrc: fix spacing when asking for fs/auth ip
ppatience0 [Sun, 28 Apr 2013 22:34:47 +0000 (18:34 -0400)]
bootrc: fix spacing when asking for fs/auth ip

11 years agofix missing return compile error for manual book index generator
cinap_lenrek [Sun, 28 Apr 2013 14:47:03 +0000 (16:47 +0200)]
fix missing return compile error for manual book index generator

11 years agomerge
ftrvxmtrx [Sat, 27 Apr 2013 22:23:20 +0000 (00:23 +0200)]
merge

11 years agoacme: fix more error messages
ftrvxmtrx [Sat, 27 Apr 2013 22:19:35 +0000 (00:19 +0200)]
acme: fix more error messages

11 years agolibdraw: avoid printing error on closemouse()
cinap_lenrek [Sat, 27 Apr 2013 02:21:08 +0000 (04:21 +0200)]
libdraw: avoid printing error on closemouse()

11 years agolibdraw: exit mouseproc() on read error instead of spinning
cinap_lenrek [Sat, 27 Apr 2013 02:10:39 +0000 (04:10 +0200)]
libdraw: exit mouseproc() on read error instead of spinning

11 years agoacme/win: fix error messages
ftrvxmtrx [Fri, 26 Apr 2013 18:35:10 +0000 (20:35 +0200)]
acme/win: fix error messages

11 years agonusb/serial: Add support for Amontec JTAGkey devices (from sources)
cinap_lenrek [Fri, 26 Apr 2013 13:25:25 +0000 (15:25 +0200)]
nusb/serial: Add support for Amontec JTAGkey devices (from sources)

11 years agofix <mach.h> types in manual (erik quanstros machmantype patch)
cinap_lenrek [Wed, 24 Apr 2013 22:13:19 +0000 (00:13 +0200)]
fix <mach.h> types in manual (erik quanstros machmantype patch)

11 years agomake all the commands agnostic about Rune width. (from sources)
cinap_lenrek [Wed, 24 Apr 2013 18:13:18 +0000 (20:13 +0200)]
make all the commands agnostic about Rune width. (from sources)

11 years agocpu: don't print spurious errstr for bad auth method, fix comment
cinap_lenrek [Wed, 24 Apr 2013 00:41:36 +0000 (02:41 +0200)]
cpu: don't print spurious errstr for bad auth method, fix comment

11 years agofortunes: i would not be at all surprised if the nix kernel (with or without the...
stanley lieber [Tue, 23 Apr 2013 22:53:37 +0000 (18:53 -0400)]
fortunes: i would not be at all surprised if the nix kernel (with or without the AC stuff) ends up being the basis for continuing evolution of plan9.

11 years agoftpd: "opts utf8 on"
cinap_lenrek [Tue, 23 Apr 2013 17:37:56 +0000 (19:37 +0200)]
ftpd: "opts utf8 on"

11 years agoftpd: fix nil pointer derefernece when ominiting arguments to opts command
cinap_lenrek [Tue, 23 Apr 2013 20:47:27 +0000 (22:47 +0200)]
ftpd: fix nil pointer derefernece when ominiting arguments to opts command

11 years agoftpd: implement "OPTS UTF-8 ON", fix uninitialized printing uninitialized memory...
cinap_lenrek [Tue, 23 Apr 2013 20:38:03 +0000 (22:38 +0200)]
ftpd: implement "OPTS UTF-8 ON", fix uninitialized printing uninitialized memory in HELP command

11 years ago- use the double-buffer buffer to allow redrawing on resize events.
cinap_lenrek [Mon, 22 Apr 2013 17:10:09 +0000 (19:10 +0200)]
- use the double-buffer buffer to allow redrawing on resize events.
specifing -d on the command line now only disables synchronous
drawing events.

- use threaded mouse and keyboard to allow for asynchronous
receoption of quit messages.  this allows plot to exit before drawing
is completed.  for programs like mapdemo, this is important.

there were two things that needed to get fixed as a result
- replace fprint(2, ...); exits("bad"); with sysfatal.  also get rid
of stdio.

- dpoint needed a mach-dependentent (sic) version.  otherwise
points on a resized screen will not be properly placed.

11 years agoether8169: fix deadlock in allocation error case for attach. (thanks erik)
cinap_lenrek [Mon, 22 Apr 2013 16:50:29 +0000 (18:50 +0200)]
ether8169: fix deadlock in allocation error case for attach. (thanks erik)

typo in error case, should qunlock() instead of qlock().

11 years agographics(2): add newwindow() key reference
cinap_lenrek [Mon, 22 Apr 2013 20:14:08 +0000 (22:14 +0200)]
graphics(2): add newwindow() key reference

11 years agoaux/listen: do not redirect stderr (fd 2) of the listener to the network connection
cinap_lenrek [Sat, 20 Apr 2013 19:42:20 +0000 (21:42 +0200)]
aux/listen: do not redirect stderr (fd 2) of the listener to the network connection

Charles Forsyth described the problem below in:

http://9fans.net/archive/2013/04/190

In a few cases, the kernel will use pprint to put a diagnostic on the
standard error (file descriptor 2). One of those is a warning that the
process has used more than 100 file descriptors. That message is possibly
obsolete and could be removed, but there are others, such as notifying an
uncaught trap that are probably helpful to make visible. In any case, as
things stand, a busy exportfs might have many file descriptors open,
provoking the diagnostic. Unfortunately, aux/listen and aux/listen1 connect
file descriptor 2 to the incoming network connection. If the connection's
protocol is not a simple, unstructured, textual one, diagnostics on the
standard error will cause confusion, in particular to devmnt.c if 9p is used.

/rc/bin/service files that start applications that run special protocols
might want to redirect file descriptor 2; alternatively, perhaps aux/listen
shouldn't redirect fd 2 by default: the few commands that do connect the remote
user to shells, or equivalent, including telnetd and sshd could dup 1 to 2
when that was sensible.

11 years agomerge
cinap_lenrek [Sat, 20 Apr 2013 20:51:22 +0000 (22:51 +0200)]
merge

11 years agoip/torrent: use NPROC processes in parallel for calculating initial piece hashes
cinap_lenrek [Sat, 20 Apr 2013 20:47:39 +0000 (22:47 +0200)]
ip/torrent: use NPROC processes in parallel for calculating initial piece hashes

11 years agomouse: fix cursor redraw to prevent screen blanking (thanks erik)
cinap_lenrek [Tue, 16 Apr 2013 16:52:49 +0000 (18:52 +0200)]
mouse: fix cursor redraw to prevent screen blanking (thanks erik)

11 years agodevmouse: remove unneeded reference to kerndate
cinap_lenrek [Sun, 14 Apr 2013 16:28:55 +0000 (18:28 +0200)]
devmouse: remove unneeded reference to kerndate

11 years agosdiahci: add pci id for 82801eb/er and c210
cinap_lenrek [Sun, 14 Apr 2013 16:13:26 +0000 (18:13 +0200)]
sdiahci: add pci id for 82801eb/er and c210

11 years agoreduce software cursor flickering
cinap_lenrek [Sun, 14 Apr 2013 14:28:54 +0000 (16:28 +0200)]
reduce software cursor flickering

the software cursor starts flickering and reacts bumby if a process
spends most of its time with drawlock acquired because the timer interrupt
thats supposed to redraw the cursor fails to acquire the lock at the time
the timer fires.

instead of trying to draw the cursor on the screen from a timer interrupt
30 times per second, devmouse now creates a process calling cursoron() and
cursoroff() when the cursor needs to be redrawn. this allows the swcursor
to schedule a redraw while holding the drawlock in swcursoravoid() and
cursoron()/cursoroff() are now able to wait for a qlock (drawlock) because
they get called from process context.

the overall responsiveness is also improved with this change as the cursor
redraw rate isnt limited to 30 times a second anymore.

11 years agofortunes: support for non-blocking i/o
stanley lieber [Sat, 13 Apr 2013 00:51:47 +0000 (20:51 -0400)]
fortunes: support for non-blocking i/o

11 years agomerge
ppatience0 [Fri, 12 Apr 2013 22:39:12 +0000 (18:39 -0400)]
merge

11 years agojpg(1): update bmp and yuv usage
ppatience0 [Fri, 12 Apr 2013 22:36:22 +0000 (18:36 -0400)]
jpg(1): update bmp and yuv usage

11 years agopci: Add Intel C216 chipset PCI support (from sources)
cinap_lenrek [Fri, 12 Apr 2013 21:07:40 +0000 (23:07 +0200)]
pci: Add Intel C216 chipset PCI support (from sources)

11 years agousbehci: align page td buffer for >16K transfers, fix isohsinit()
cinap_lenrek [Fri, 12 Apr 2013 20:14:32 +0000 (22:14 +0200)]
usbehci: align page td buffer for >16K transfers, fix isohsinit()

from ehci spec:

The buffer pointer list in the qTD is long enough to support a maximum
transfer size of 20K bytes. This case occurs when all five buffer pointers
are used and the first offset is zero. A qTD handles a 16Kbyte buffer
with any starting buffer alignment.

11 years agomerge
cinap_lenrek [Thu, 11 Apr 2013 17:34:43 +0000 (19:34 +0200)]
merge

11 years agondb/dns: dont override req->aborttime in udpquery()
cinap_lenrek [Thu, 11 Apr 2013 17:33:07 +0000 (19:33 +0200)]
ndb/dns: dont override req->aborttime in udpquery()

overriding aborttime in udpquery() makes no sense. it causes
recursive queries to extend the timeout infinitely. nobody
but the issuer of the request should modify aborttime.

11 years agolibauth: auth_chuid empty (from sources)
cinap_lenrek [Thu, 11 Apr 2013 11:45:11 +0000 (13:45 +0200)]
libauth: auth_chuid empty (from sources)

instead of an "i/o count too small", detect a missing capability (empty, null string)
before the write, and diagnose it as such.

11 years ago9nfs: apply patch/nfswarnandflags (from sources, thanks nemo)
cinap_lenrek [Wed, 10 Apr 2013 18:51:39 +0000 (20:51 +0200)]
9nfs: apply patch/nfswarnandflags (from sources, thanks nemo)

Remove a warning in 9nfs and make it use the std. CFLAGS
(we detected the warning using -W).

11 years agopostscript: remove private copy of utf8 implementation (from sources)
cinap_lenrek [Wed, 10 Apr 2013 18:47:01 +0000 (20:47 +0200)]
postscript: remove private copy of utf8 implementation (from sources)

11 years agoetheriwl: add pci id for Centrino Wireless-N 100 (thanks fgudin9)
cinap_lenrek [Wed, 10 Apr 2013 20:13:54 +0000 (22:13 +0200)]
etheriwl: add pci id for Centrino Wireless-N 100 (thanks fgudin9)

11 years agoape: inet_ntop() print v4 mapped addresses in v6 as dotted quad
cinap_lenrek [Fri, 5 Apr 2013 12:09:41 +0000 (14:09 +0200)]
ape: inet_ntop() print v4 mapped addresses in v6 as dotted quad

11 years agosam(1): document -i flag
ppatience0 [Thu, 4 Apr 2013 20:44:23 +0000 (16:44 -0400)]
sam(1): document -i flag

11 years agosam, samterm: add -i flag to make tab insert spaces up to the next tabstop and backsp...
ppatience0 [Thu, 4 Apr 2013 20:38:36 +0000 (16:38 -0400)]
sam, samterm: add -i flag to make tab insert spaces up to the next tabstop and backspace to delete spaces until the previous tabstop

11 years agointel ethernet: default to 16 byte cache line size when not properly initialized...
cinap_lenrek [Wed, 3 Apr 2013 19:51:42 +0000 (21:51 +0200)]
intel ethernet: default to 16 byte cache line size when not properly initialized and disable checksum offload for igbe (from sources)

11 years agopci: add pci bridge id for Intel 6 Series/C200 (from sources)
cinap_lenrek [Wed, 3 Apr 2013 19:47:50 +0000 (21:47 +0200)]
pci: add pci bridge id for Intel 6 Series/C200 (from sources)

11 years agopushtls: remove unneccesary include <auth.h> (from sources)
cinap_lenrek [Wed, 3 Apr 2013 19:46:33 +0000 (21:46 +0200)]
pushtls: remove unneccesary include <auth.h> (from sources)

11 years agoadd import -z option to skip initial tree negotiation (from mycroftiv)
cinap_lenrek [Tue, 2 Apr 2013 02:23:26 +0000 (04:23 +0200)]
add import -z option to skip initial tree negotiation (from mycroftiv)

11 years agoape: inet_pton() parse dotted address to IPv4 mapped addresses for AF_INET6
cinap_lenrek [Tue, 2 Apr 2013 01:48:10 +0000 (03:48 +0200)]
ape: inet_pton() parse dotted address to IPv4 mapped addresses for AF_INET6

11 years agofix parseip()
cinap_lenrek [Tue, 2 Apr 2013 01:39:24 +0000 (03:39 +0200)]
fix parseip()

addresses like: "1:2:3:4:5:6:7:255.255.255.255" caused parseip
to write beyond the ip buffer.

11 years agoape: check *alen before copying in getpeername(), getsockname() and accept()
cinap_lenrek [Mon, 1 Apr 2013 23:40:29 +0000 (01:40 +0200)]
ape: check *alen before copying in getpeername(), getsockname() and accept()

*alen has to be initialized to the size of the buffer
by the caller, and we are supposed to put the real
size of the address in there, but not copy more than
the original *alen value (truncate).

11 years agoape: add inet_aton()
cinap_lenrek [Mon, 1 Apr 2013 17:16:18 +0000 (19:16 +0200)]
ape: add inet_aton()

11 years agopython: update python build configuration to new ape capabilities like getaddrinfo...
cinap_lenrek [Sun, 31 Mar 2013 16:54:45 +0000 (18:54 +0200)]
python: update python build configuration to new ape capabilities like getaddrinfo(), ipv6 and fix EISCON typo in ape

11 years agoape: initial IPv6 support, inet_pton()/inet_ntop(), getaddrinfo()/getnameinfo()
cinap_lenrek [Sun, 31 Mar 2013 16:52:45 +0000 (18:52 +0200)]
ape: initial IPv6 support, inet_pton()/inet_ntop(), getaddrinfo()/getnameinfo()

11 years agoape: define FD_SETSIZE for select()
cinap_lenrek [Sun, 31 Mar 2013 16:46:14 +0000 (18:46 +0200)]
ape: define FD_SETSIZE for select()

11 years agomerge
ppatience0 [Sun, 31 Mar 2013 02:52:30 +0000 (22:52 -0400)]
merge

11 years agoether82563: add untested devices from OpenBSD's em(4) Intel driver. also update some...
ppatience0 [Sun, 31 Mar 2013 02:50:48 +0000 (22:50 -0400)]
ether82563: add untested devices from OpenBSD's em(4) Intel driver. also update some names according to OpenBSD's driver (previously added device, 0x10c0, is also from OpenBSD's driver).

11 years agoape: fix potential double close in socket() and protect rock chain with spinlock
cinap_lenrek [Fri, 29 Mar 2013 19:49:49 +0000 (20:49 +0100)]
ape: fix potential double close in socket() and protect rock chain with spinlock

11 years agoether82563: add 82562V-2 did
ppatience0 [Sun, 24 Mar 2013 05:51:43 +0000 (01:51 -0400)]
ether82563: add 82562V-2 did

11 years agofortunes: Now reboot twice[1].
stanley lieber [Sun, 24 Mar 2013 23:49:02 +0000 (19:49 -0400)]
fortunes: Now reboot twice[1].

11 years agowifi: cleanup
cinap_lenrek [Sun, 24 Mar 2013 20:59:47 +0000 (21:59 +0100)]
wifi: cleanup

cleanup wifiencrypt()/wifidecrypt() routines moving tkip
and ccmp in separate routines so we dont use less stack
space.

11 years agowpa(8) document -1 and -2 options
cinap_lenrek [Sun, 24 Mar 2013 01:30:11 +0000 (02:30 +0100)]
wpa(8) document -1 and -2 options

11 years agowpa: fix rsc for wpa2
cinap_lenrek [Sat, 23 Mar 2013 21:09:46 +0000 (22:09 +0100)]
wpa: fix rsc for wpa2

in wpa2, the rsc field of the eapol message3 is the packet number for
*group* messages that the ap will use as there is no separate group
key message. in wpa1, we use it for the peerwise key.

11 years agounrolling loops in aesCCMencrypt() and aesCCMdecrypt()
cinap_lenrek [Sat, 23 Mar 2013 19:52:54 +0000 (20:52 +0100)]
unrolling loops in aesCCMencrypt() and aesCCMdecrypt()

do the xoring word wise for 16-byte block instead
of doing it bytewise in a loop.

11 years agowpa: remove debug print
cinap_lenrek [Sat, 23 Mar 2013 03:50:43 +0000 (04:50 +0100)]
wpa: remove debug print

11 years agomerge
cinap_lenrek [Sat, 23 Mar 2013 03:31:45 +0000 (04:31 +0100)]
merge

11 years agowpa2 support continued...
cinap_lenrek [Sat, 23 Mar 2013 03:30:20 +0000 (04:30 +0100)]
wpa2 support continued...

11 years agoadd experimental wpa2/ccmp support
cinap_lenrek [Sat, 23 Mar 2013 03:27:49 +0000 (04:27 +0100)]
add experimental wpa2/ccmp support

11 years agofortunes: stockholm
khm [Fri, 22 Mar 2013 23:25:50 +0000 (19:25 -0400)]
fortunes: stockholm

11 years agomerge
cinap_lenrek [Sun, 17 Mar 2013 18:41:39 +0000 (19:41 +0100)]
merge

11 years agowpa: initial support for rsn
cinap_lenrek [Sun, 17 Mar 2013 18:39:44 +0000 (19:39 +0100)]
wpa: initial support for rsn

11 years agoaux/cpuid: xsaveopt
ftrvxmtrx [Sun, 17 Mar 2013 12:20:03 +0000 (13:20 +0100)]
aux/cpuid: xsaveopt

11 years agoaux/cpuid: better looking arrays
ftrvxmtrx [Sat, 16 Mar 2013 20:13:49 +0000 (21:13 +0100)]
aux/cpuid: better looking arrays

11 years agoaux/cpuid: fix wrong extfunc1 bits
ftrvxmtrx [Sat, 16 Mar 2013 17:07:48 +0000 (18:07 +0100)]
aux/cpuid: fix wrong extfunc1 bits

11 years agosysexec: fix possible segment overlap with temporary stack
cinap_lenrek [Sat, 16 Mar 2013 01:37:07 +0000 (02:37 +0100)]
sysexec: fix possible segment overlap with temporary stack

the kernel uses fixed area (TSTKTOP, TSTKSIZ) of the address
space to temporarily map the new stack segment for exec. for
386 and arm, this area was right below the stack segment which
has the problem that the program can map arbitrary segments
there (even readonly).

alpha and ppc dont have this problem as they map the temporary
exec stack *above* the user reachable stack segement and segattach
prevents one from mapping anything above or overlaping the stack.

lots of arch code assumes USTKTOP being the end of userspace
address space and changing this to TSTKTOP would work, but results
in lots of hard to test changes.

instead, we'r going to map the temporary stack programmatically
finding a hole in the address space where to map it. we also lift
the size limitation for arguments and allow arguments to fill
the whole new stack segement.

the TSTKTOP and TSTKSIZ are not used anymore so they where removed.

references:

http://9fans.net/archive/2013/03/203
http://9fans.net/archive/2013/03/202
http://9fans.net/archive/2013/03/197
http://9fans.net/archive/2013/03/195
http://9fans.net/archive/2013/03/181

11 years agomerge
cinap_lenrek [Thu, 14 Mar 2013 19:56:30 +0000 (20:56 +0100)]
merge

11 years agohjfs: fix the megashit
cinap_lenrek [Thu, 14 Mar 2013 19:55:16 +0000 (20:55 +0100)]
hjfs: fix the megashit

aiju → what is this huge if(d != nil) {} megashit

if getdent() fails in newentry() then return immidiately
and dont override the error string.

11 years agoinst/mountcwfs, inst/mounthjfs: fix typos
ppatience0 [Tue, 12 Mar 2013 21:55:00 +0000 (17:55 -0400)]
inst/mountcwfs, inst/mounthjfs: fix typos

11 years agomerge
cinap_lenrek [Tue, 12 Mar 2013 20:08:47 +0000 (21:08 +0100)]
merge

11 years agoether8169: add RTL8101/8102E macid
cinap_lenrek [Tue, 12 Mar 2013 20:07:40 +0000 (21:07 +0100)]
ether8169: add RTL8101/8102E macid

11 years agoetheriwl: 6005 - set rom calibration flags and temperature sensor offset
ftrvxmtrx [Mon, 11 Mar 2013 23:14:55 +0000 (00:14 +0100)]
etheriwl: 6005 - set rom calibration flags and temperature sensor offset

11 years agoape: threadsafe errno
cinap_lenrek [Mon, 11 Mar 2013 17:57:22 +0000 (18:57 +0100)]
ape: threadsafe errno

store errno on the private process stack so its always per process
and not just per memory space. errno itself becomes a macro
dereferencing int *_errnoloc; which is initialized from main9.s
pointing to the private stack location.

various fixes in programs that just imported errno variable with
"extern int errno;" instead of including <errno.h>.

11 years agomerge
cinap_lenrek [Mon, 11 Mar 2013 00:17:45 +0000 (01:17 +0100)]
merge

11 years agoape: fix thread race with close() and select()
cinap_lenrek [Mon, 11 Mar 2013 00:16:34 +0000 (01:16 +0100)]
ape: fix thread race with close() and select()

in ape close(), do the real filedescriptor _CLOSE() *after* we cleared
the _fdinfo[] slot because once closed, we dont own the slot anymore and
another process doing open() can trash the slot. make sure open() retuns
fd < OPEN_MAX.

double check in _startbuf() holding mux->lock if the fd is already buffered
preveting running double copyprocs on a fd.

dont zero the mux->rwant/ewant bitmaps at the end of select() as we do not
hold the mix->lock.

in _closebuf() kill copyproc while holding the mux->lock to make sure the
copyproc isnt holding it at the time it is killed. run kill() multiple times
to make sure the proc is gone.

11 years agoape/malloc: make malloc and free threadsafe for python
cinap_lenrek [Sun, 10 Mar 2013 23:55:26 +0000 (00:55 +0100)]
ape/malloc: make malloc and free threadsafe for python

11 years agoape/stdio: make fopen() quasi threadsafe for python
cinap_lenrek [Sun, 10 Mar 2013 23:48:35 +0000 (00:48 +0100)]
ape/stdio: make fopen() quasi threadsafe for python

python uses processes sharing memory. it requires at least fopen() to
be called by multiple threads at once so we introduce _IO_newfile()
which allocates the FILE structure slot under a lock.

11 years agowpa(8): fix typos
ppatience0 [Sun, 10 Mar 2013 23:25:01 +0000 (19:25 -0400)]
wpa(8): fix typos

11 years agoape: check for invalid filedescriptors in select()
cinap_lenrek [Sun, 10 Mar 2013 17:35:07 +0000 (18:35 +0100)]
ape: check for invalid filedescriptors in select()

11 years agonusb/disk: add small delay after unit start command (for thinkpad sdcard reader)
cinap_lenrek [Sun, 10 Mar 2013 03:06:42 +0000 (04:06 +0100)]
nusb/disk: add small delay after unit start command (for thinkpad sdcard reader)

diskparts fails on thinkpad x200 sdcard reader if we dont
give the unit some time to startup. the device took about
100ms to become ready so status poll loop with a long
timeout would be overkill.

11 years agoprovide wpa(8) manual page
cinap_lenrek [Sat, 9 Mar 2013 18:18:00 +0000 (19:18 +0100)]
provide wpa(8) manual page

11 years agoaux/wpa: check reply counter only after mic check
cinap_lenrek [Sat, 9 Mar 2013 17:12:49 +0000 (18:12 +0100)]
aux/wpa: check reply counter only after mic check