]> git.lizzy.rs Git - plan9front.git/log
plan9front.git
8 years agolibsec: add chacha cipher (from charles forsyth)
cinap_lenrek [Thu, 22 Oct 2015 05:48:26 +0000 (07:48 +0200)]
libsec: add chacha cipher (from charles forsyth)

8 years agolibsec: add poly1305
cinap_lenrek [Thu, 22 Oct 2015 05:17:25 +0000 (07:17 +0200)]
libsec: add poly1305

8 years agofortunes: Maybe a git branch can help here?
stanley lieber [Mon, 19 Oct 2015 15:56:03 +0000 (11:56 -0400)]
fortunes: Maybe a git branch can help here?

8 years agoacid/leak: fix endless loop for B2NB(b) == b case (thanks mischief)
cinap_lenrek [Sun, 18 Oct 2015 22:20:33 +0000 (00:20 +0200)]
acid/leak: fix endless loop for B2NB(b) == b case (thanks mischief)

8 years agowifi: prioritize rsne over wpaie
cinap_lenrek [Sat, 17 Oct 2015 12:45:25 +0000 (14:45 +0200)]
wifi: prioritize rsne over wpaie

if beacon/proble contains both wpa information element and rsne, the rsne wins.

8 years agomerge
cinap_lenrek [Thu, 15 Oct 2015 11:22:48 +0000 (13:22 +0200)]
merge

8 years agolib9p: do not override Srv.end in listensrv(), simplify srvclose() and recounting
cinap_lenrek [Thu, 15 Oct 2015 11:21:30 +0000 (13:21 +0200)]
lib9p: do not override Srv.end in listensrv(), simplify srvclose() and recounting

listensrv() used to override Srv.end() with its own handler
to free the malloc'd Srv structure and close the fd. this
makes it impossible to register your own cleanup handler.
instead, we introduce the private Srv.free() handler that
is used by listensrv to register its cleanup code. Srv.free()
is called once all the srv procs have been exited and all
requests on that srv have been responded to while Srv.end()
is called once all the procs exited the srv loop regardless
of the requests still being in flight.

8 years agolib9p: add reqqueuefree
aiju [Mon, 12 Oct 2015 08:41:40 +0000 (10:41 +0200)]
lib9p: add reqqueuefree

8 years agolibsec: remove rc4 cipher suits for tls
cinap_lenrek [Fri, 9 Oct 2015 23:12:30 +0000 (01:12 +0200)]
libsec: remove rc4 cipher suits for tls

8 years agoaux/listen1: usage()
cinap_lenrek [Fri, 9 Oct 2015 22:11:03 +0000 (00:11 +0200)]
aux/listen1: usage()

8 years agoaux/listen1: allow alternative namespace when running as user none with -n option
cinap_lenrek [Fri, 9 Oct 2015 22:09:02 +0000 (00:09 +0200)]
aux/listen1: allow alternative namespace when running as user none with -n option

8 years agocwfs: fix wstat() failing to mark block dirty when noatime is set
cinap_lenrek [Fri, 9 Oct 2015 17:22:53 +0000 (19:22 +0200)]
cwfs: fix wstat() failing to mark block dirty when noatime is set

code assumed the accessdir() call would always mark the block dirty, but
this is not the case when noatime flag is enabled. this was reported by
michael in bug:

"open/with_noatime_option_cwfs_doesnt_preserve_changes_in_file_permissionowner"

--
cinap

8 years agolibsec: clarify the code by declaring constant for DirectoryString and handle convers...
cinap_lenrek [Fri, 9 Oct 2015 16:52:32 +0000 (18:52 +0200)]
libsec: clarify the code by declaring constant for DirectoryString and handle conversion in mkstring()

8 years agolibsec: use the right string encoding for various x509 fields
cinap_lenrek [Fri, 9 Oct 2015 16:31:49 +0000 (18:31 +0200)]
libsec: use the right string encoding for various x509 fields

C= has to be PrintableString, E= and DC= are IA5String, everything
else is DirectoryString which can be PrintableString or UTF8String.

8 years agowifi: quote value of parsed ether options
cinap_lenrek [Fri, 9 Oct 2015 03:10:47 +0000 (05:10 +0200)]
wifi: quote value of parsed ether options

introduce wificfg() function to convert ether->opt[] strings
to wifictl messages, which needs quoting for the value. so
etherX=type=iwl essid='something with spaces' works.

8 years agomerge
cinap_lenrek [Thu, 8 Oct 2015 15:16:06 +0000 (17:16 +0200)]
merge

8 years agocc: restore side(), but do not consider OINDEX as side effect free
cinap_lenrek [Thu, 8 Oct 2015 15:14:07 +0000 (17:14 +0200)]
cc: restore side(), but do not consider OINDEX as side effect free

from charles forsuth:

because the previous version thought OINDEX might have a side effect, it
stopped it building a tower of them.
probably the best thing is to limit that anyway, since each one consumes
2-3 registers, so towering them can
keep even more active, and the x86 hasn't got that many.

the quick hack is to return that case to the earlier state by treating
OINDEX as a side-effect in side().
it's not a bad thing to do in the OSTRUCT case, for similar reasons: it's
better to collapse the indexed pointer
into a direct register, instead of repeating the indexing operation through
the copying of the value.
OINDEX isn't a machine-independent operation, so it doesn't affect the uses
in ../cc

8 years agoport: fix typo in devmnt mntproc name
mischief [Thu, 8 Oct 2015 04:45:03 +0000 (21:45 -0700)]
port: fix typo in devmnt mntproc name

8 years agodoom: fix bogus pointer arithmetic, do sfx initialization in two passes for linked...
cinap_lenrek [Wed, 7 Oct 2015 22:37:54 +0000 (00:37 +0200)]
doom: fix bogus pointer arithmetic, do sfx initialization in two passes for linked sounds

8 years agocc: back out side() change unil 8c is fixed
cinap_lenrek [Wed, 7 Oct 2015 11:35:36 +0000 (13:35 +0200)]
cc: back out side() change unil 8c is fixed

this change made 8c fail to build libvorbis and gs, until this
is fixed we can live with non-optimal code.

8 years agoqc: import changes from charles forsyth
cinap_lenrek [Wed, 7 Oct 2015 07:40:48 +0000 (09:40 +0200)]
qc: import changes from charles forsyth

8 years agokc: import various changes from charles forsyth
cinap_lenrek [Tue, 6 Oct 2015 04:55:05 +0000 (06:55 +0200)]
kc: import various changes from charles forsyth

8 years agocc: include $builtin as keywoard in pickle() (from charles forsyth)
cinap_lenrek [Tue, 6 Oct 2015 04:47:55 +0000 (06:47 +0200)]
cc: include $builtin as keywoard in pickle() (from charles forsyth)

8 years agocc: getflag() fmt will be terminated at end of loop (from charles forsyth)
cinap_lenrek [Tue, 6 Oct 2015 04:38:37 +0000 (06:38 +0200)]
cc: getflag() fmt will be terminated at end of loop (from charles forsyth)

8 years agocc: allow runes as macro names (from charles forsyth)
cinap_lenrek [Tue, 6 Oct 2015 04:34:30 +0000 (06:34 +0200)]
cc: allow runes as macro names (from charles forsyth)

8 years ago5c/6c/8c/vc: import various changes from charles forsyth
cinap_lenrek [Tue, 6 Oct 2015 04:20:01 +0000 (06:20 +0200)]
5c/6c/8c/vc: import various changes from charles forsyth

- cover more cases that have no side effects
- ensure function has complex FNX
- pull operators out of OFUNC level
- rewrite OSTRUCT lhs to avoid all side-effects, use regalloc() instead of regret()

8 years agocc: TUSHORT -> TRUNE for lstring constats for OUSED
cinap_lenrek [Tue, 6 Oct 2015 03:20:00 +0000 (05:20 +0200)]
cc: TUSHORT -> TRUNE for lstring constats for OUSED

8 years ago9660srv: do not ignore upper 32 bits of offset when reading directory
cinap_lenrek [Mon, 5 Oct 2015 02:45:32 +0000 (04:45 +0200)]
9660srv: do not ignore upper 32 bits of offset when reading directory

8 years agodossrv: handle file offsets > 2^31
cinap_lenrek [Mon, 5 Oct 2015 02:31:31 +0000 (04:31 +0200)]
dossrv: handle file offsets > 2^31

maximum file size is 4GB-1 as the file length is stored in
a 32 bit long. make sure it doesnt overflow on write or
or truncate. interpret the file length as unsigned. pass
vlong to readfile()/writefile()/truncfile() so we can
handle overflows and not just ignore the upper bits.

8 years ago6c: remove 6c/vlrt.c file
cinap_lenrek [Sun, 4 Oct 2015 20:12:16 +0000 (22:12 +0200)]
6c: remove 6c/vlrt.c file

8 years ago8c: dont abort() when running out of registers.
cinap_lenrek [Sun, 4 Oct 2015 20:09:36 +0000 (22:09 +0200)]
8c: dont abort() when running out of registers.

8 years agovc: fix mistake.
cinap_lenrek [Sun, 4 Oct 2015 19:22:17 +0000 (21:22 +0200)]
vc: fix mistake.

8 years agoape: add missing _subv() function to 386/vlop.s
cinap_lenrek [Sun, 4 Oct 2015 18:16:41 +0000 (20:16 +0200)]
ape: add missing _subv() function to 386/vlop.s

8 years agovc: handle 64 bit mixedmode asop
cinap_lenrek [Sun, 4 Oct 2015 18:08:10 +0000 (20:08 +0200)]
vc: handle 64 bit mixedmode asop

8 years ago5c: handle 64 bit mixedmode asop
cinap_lenrek [Sun, 4 Oct 2015 18:07:34 +0000 (20:07 +0200)]
5c: handle 64 bit mixedmode asop

8 years ago8c: handle 64 bit mixedmode asop and type vlong <-> float/double type conversions
cinap_lenrek [Sun, 4 Oct 2015 18:06:59 +0000 (20:06 +0200)]
8c: handle 64 bit mixedmode asop and type vlong <-> float/double type conversions

8 years agocc: handle 64 bit mixedmode asop and type vlong <-> float/double type conversions
cinap_lenrek [Sun, 4 Oct 2015 18:06:30 +0000 (20:06 +0200)]
cc: handle 64 bit mixedmode asop and type vlong <-> float/double type conversions

8 years agocc/6c: fix return type of mixed asop expressions, preserve type for moves so fixed...
cinap_lenrek [Sun, 4 Oct 2015 18:02:58 +0000 (20:02 +0200)]
cc/6c: fix return type of mixed asop expressions, preserve type for moves so fixed<->float conversions work correctly

8 years agoape: add machine specific code for spim
cinap_lenrek [Sun, 4 Oct 2015 17:50:24 +0000 (19:50 +0200)]
ape: add machine specific code for spim

8 years agolibc: add _uv2d()/uv2f() and _vas*d() functions to vlrt.c
cinap_lenrek [Sun, 4 Oct 2015 17:49:02 +0000 (19:49 +0200)]
libc: add _uv2d()/uv2f() and _vas*d() functions to vlrt.c

on 32 bit archs, implement 64 bit vasop with floatingpoint right hand side.
also added is uvlong->double conversion function.

8 years agomerge
cinap_lenrek [Sat, 3 Oct 2015 10:19:31 +0000 (12:19 +0200)]
merge

8 years ago8c: make cgen64() compile target first when it contains functoin call so final assign...
cinap_lenrek [Sat, 3 Oct 2015 10:18:20 +0000 (12:18 +0200)]
8c: make cgen64() compile target first when it contains functoin call so final assignment wont trash the registers

8 years agopc, pc64: add did for intel wildcat point audio controller
mischief [Wed, 30 Sep 2015 01:44:30 +0000 (18:44 -0700)]
pc, pc64: add did for intel wildcat point audio controller

8 years agoadd timepic(1) and qr(1) manpages
aiju [Tue, 29 Sep 2015 18:49:52 +0000 (20:49 +0200)]
add timepic(1) and qr(1) manpages

8 years agotimepic: string support
aiju [Tue, 29 Sep 2015 18:49:28 +0000 (20:49 +0200)]
timepic: string support

8 years agoape: set MB_CUR_MAX to 4 in stdlib.h for 21-bit runes (thanks erik quanstro)
cinap_lenrek [Tue, 29 Sep 2015 13:17:21 +0000 (15:17 +0200)]
ape: set MB_CUR_MAX to 4 in stdlib.h for 21-bit runes (thanks erik quanstro)

> fix silly bug with ape/stdlib.h which caused ape/lib/regexp/regcomp.c
> to miscompile regular expressions with runes > 0xffff.

8 years agoaux/gpsfs: fix usage
BurnZeZ [Tue, 29 Sep 2015 08:08:42 +0000 (04:08 -0400)]
aux/gpsfs: fix usage

8 years agofile: detect DICOM and XM audio headers
BurnZeZ [Mon, 28 Sep 2015 07:42:26 +0000 (03:42 -0400)]
file: detect DICOM and XM audio headers

8 years agoteach doctype how to timepic
aiju [Mon, 28 Sep 2015 09:33:06 +0000 (11:33 +0200)]
teach doctype how to timepic

8 years agodevip: declare cleanarpent() static
cinap_lenrek [Sun, 27 Sep 2015 20:41:38 +0000 (22:41 +0200)]
devip: declare cleanarpent() static

8 years agodevip: various bugfixes and cleanups for arp code
cinap_lenrek [Sun, 27 Sep 2015 20:17:02 +0000 (22:17 +0200)]
devip: various bugfixes and cleanups for arp code

- fix missing runlock(ifc) when ifcid != a->ifcid in rxmitsols() (thanks erik quanstro)
- don't leak packets when transfering blocks from arp entry hold list to droplist
- free rest of droplist when bwrite() errors in arpenter(), remove useless checks (ifc != nil)
- free arp entry hold list from cleanarpent()
- consistent use of nil for pointers

8 years agoadd timepic(1)
aiju [Sun, 27 Sep 2015 11:50:18 +0000 (13:50 +0200)]
add timepic(1)

8 years agofortunes: Please be respectful in this forum.
stanley lieber [Sat, 26 Sep 2015 23:10:40 +0000 (19:10 -0400)]
fortunes: Please be respectful in this forum.

8 years agoqer(8): correct man page example (thanks, kenji)
stanley lieber [Sat, 26 Sep 2015 23:07:05 +0000 (19:07 -0400)]
qer(8): correct man page example (thanks, kenji)

8 years agomerge
cinap_lenrek [Sat, 26 Sep 2015 09:46:31 +0000 (11:46 +0200)]
merge

8 years agocwfs: various cleanups
cinap_lenrek [Sat, 26 Sep 2015 09:44:29 +0000 (11:44 +0200)]
cwfs: various cleanups

- consistently refer to Dentry.name length with NAMELEN
- make sure whoname is null terminated
- remove useless nil check for whoname

8 years agocwfs: don't use sprint() to fill directory name
cinap_lenrek [Sat, 26 Sep 2015 09:41:31 +0000 (11:41 +0200)]
cwfs: don't use sprint() to fill directory name

sprint() will replace invalid utf8 sequences with U+FFFD
which caused directory reads and stats to return the wrong
filename. just strcpy the name bytes.

8 years agoape: add badrect object into ape libdraw
mischief [Fri, 25 Sep 2015 23:44:25 +0000 (16:44 -0700)]
ape: add badrect object into ape libdraw

8 years agolibsec: cleanup newbytes()/newints()/newbits() and get rid of OFFSETOF() macro
cinap_lenrek [Thu, 24 Sep 2015 10:57:05 +0000 (12:57 +0200)]
libsec: cleanup newbytes()/newints()/newbits() and get rid of OFFSETOF() macro

8 years agocpp: handle 4 byte utf sequences (21-bit runes)
cinap_lenrek [Thu, 24 Sep 2015 10:23:17 +0000 (12:23 +0200)]
cpp: handle 4 byte utf sequences (21-bit runes)

8 years agoutf(6), rune(2): document 21-bit runes
cinap_lenrek [Thu, 24 Sep 2015 10:14:08 +0000 (12:14 +0200)]
utf(6), rune(2): document 21-bit runes

8 years agocpp: fix memory corruption due to input buffer relocation
cinap_lenrek [Thu, 24 Sep 2015 03:13:03 +0000 (05:13 +0200)]
cpp: fix memory corruption due to input buffer relocation

the dynamic input buffer resize code (fillbuf()) is broken as
the calling code assumes that memory wont relocate. instead
of trying to work out all the cases where this happens, i'm
getting rid of fillbuf() and just read the whole file into
memory in setsource().

the bug could be reproduced with something as simple as:

@{for(i in `{seq 1 10000}){echo $i ', \'; }} | cpp

8 years agoresize: add -n for nearest neighbour
glenda [Wed, 23 Sep 2015 15:52:35 +0000 (17:52 +0200)]
resize: add -n for nearest neighbour

8 years agomerge
cinap_lenrek [Wed, 23 Sep 2015 15:35:16 +0000 (17:35 +0200)]
merge

8 years agolibsec: fix memory leaks in X509req() and X509gen() and return malloced pointer
cinap_lenrek [Wed, 23 Sep 2015 15:33:52 +0000 (17:33 +0200)]
libsec: fix memory leaks in X509req() and X509gen() and return malloced pointer

X509req() and X509gen() used to leak memory, and had no way for
the caller to free the allocated certificate/certificate request
buffer returned. this is not critical as these functions are only
used in short lived rsa(2) helper programs. but i prefer to have
library routines not leak memory as one does not know in advance
where the code is going to be used.

8 years agorsa2x509, rsa2csr: add newline in usage print
cinap_lenrek [Wed, 23 Sep 2015 15:05:48 +0000 (17:05 +0200)]
rsa2x509, rsa2csr: add newline in usage print

8 years agolibsec: implement client certificate authentication for tls1.2
cinap_lenrek [Wed, 23 Sep 2015 14:57:25 +0000 (16:57 +0200)]
libsec: implement client certificate authentication for tls1.2

we used to negotiate tls1.1 for client cert authentication because the
signature generation was not implemented for tls1.2. this is now fixed
and tls1.2 can be negotiated with client certs.

8 years agoadd qr.c
glenda [Wed, 23 Sep 2015 14:36:55 +0000 (16:36 +0200)]
add qr.c

8 years agolibsec: handle missing signature case; can happen because some ciphers make it optional
cinap_lenrek [Tue, 22 Sep 2015 17:11:54 +0000 (19:11 +0200)]
libsec: handle missing signature case; can happen because some ciphers make it optional

8 years agolibsec: make sure Elem is zero initialized so freevalfields() wont cause accidents
cinap_lenrek [Tue, 22 Sep 2015 16:39:48 +0000 (18:39 +0200)]
libsec: make sure Elem is zero initialized so freevalfields() wont cause accidents

8 years agolibsec: implement dh parameter signature verification, stop lying about non-rsa ciphe...
cinap_lenrek [Tue, 22 Sep 2015 16:10:52 +0000 (18:10 +0200)]
libsec: implement dh parameter signature verification, stop lying about non-rsa ciphers, fix memory leaks in X509 code

actually verify the diffie hellman parameter signature, this
comes in two flavours. TLS1.2 uses X509 signature with a
single hash specified by the signature algorithm field in
the signature itself and pre TLS1.2 where md5+sha1 hashes
of the signed blob are pkcs1 padded and encrypted with the
rsa private key.

stop advertizing non-rsa cipher suits (DSS and ECDSA), as
we have not implmenented them.

fix some memory leaks in X509 code while we'r at it.

8 years agolibsec: fix memory leak in ecmul()
cinap_lenrek [Tue, 22 Sep 2015 16:04:06 +0000 (18:04 +0200)]
libsec: fix memory leak in ecmul()

8 years agoscuzz: uncomment synccache command
cinap_lenrek [Tue, 22 Sep 2015 10:23:49 +0000 (12:23 +0200)]
scuzz: uncomment synccache command

8 years agomothra: <hr> drawing
cinap_lenrek [Tue, 22 Sep 2015 09:49:07 +0000 (11:49 +0200)]
mothra: <hr> drawing

<hr> is handled by drawing replicated bitmap across the with of
the page.

8 years agolibdraw: remove unneeded check (thanks BurnZeZ)
cinap_lenrek [Sun, 20 Sep 2015 19:52:47 +0000 (21:52 +0200)]
libdraw: remove unneeded check (thanks BurnZeZ)

8 years agodevsd: handle SYNCHRONIZE CACHE scsi commands as nops in sdfakescsi()
cinap_lenrek [Sun, 20 Sep 2015 12:54:49 +0000 (14:54 +0200)]
devsd: handle SYNCHRONIZE CACHE scsi commands as nops in sdfakescsi()

8 years agosdmmc: handle fakescsi emulation
cinap_lenrek [Sun, 20 Sep 2015 12:53:44 +0000 (14:53 +0200)]
sdmmc: handle fakescsi emulation

8 years agodevsd: remove unused timeout field from SDreq
cinap_lenrek [Sun, 20 Sep 2015 12:27:41 +0000 (14:27 +0200)]
devsd: remove unused timeout field from SDreq

8 years agolibdraw, screenrc: bind devdraw and devmouse in screenrc instead of handling it in...
cinap_lenrek [Sun, 20 Sep 2015 10:25:01 +0000 (12:25 +0200)]
libdraw, screenrc: bind devdraw and devmouse in screenrc instead of handling it in libdraw

libdraw was attempting to bind '#i' and '#m' to /dev when it could not find
/dev/mouse or /dev/draw. a library shouldnt be that clever and do namespace
manipulations on behalf of the caller. so instead, we setup the graphics
environment in screenrc on boot time.

8 years agopc, pc64: fix sdvirtio descriptor count when sending flush
mischief [Sun, 20 Sep 2015 08:56:56 +0000 (01:56 -0700)]
pc, pc64: fix sdvirtio descriptor count when sending flush

8 years agoircrc: add -P option for server password (thanks, nick)
stanley lieber [Sat, 19 Sep 2015 18:03:22 +0000 (14:03 -0400)]
ircrc: add -P option for server password (thanks, nick)

8 years ago/sys/lib/dist/ndb/common: correct authdom=inri
stanley lieber [Wed, 16 Sep 2015 17:29:04 +0000 (13:29 -0400)]
/sys/lib/dist/ndb/common: correct authdom=inri

8 years agoigfx: fix typos (thanks qwx)
cinap_lenrek [Mon, 14 Sep 2015 13:15:14 +0000 (15:15 +0200)]
igfx: fix typos (thanks qwx)

8 years agolibsec: save some space making weakCipher bitamp of type char[] instead of int[]
cinap_lenrek [Mon, 14 Sep 2015 07:46:40 +0000 (09:46 +0200)]
libsec: save some space making weakCipher bitamp of type char[] instead of int[]

8 years agolibsec: add TLS_ECDHE_ECDSA_WITH_AES_128_CBC_SHA256 and TLS_ECDHE_RSA_WITH_AES_128_CB...
cinap_lenrek [Mon, 14 Sep 2015 07:27:06 +0000 (09:27 +0200)]
libsec: add TLS_ECDHE_ECDSA_WITH_AES_128_CBC_SHA256 and TLS_ECDHE_RSA_WITH_AES_128_CBC_SHA256 cipher suits

8 years agolibfis: fix wrong shift for lba40 (thanks geoff and charles)
cinap_lenrek [Sun, 13 Sep 2015 11:51:00 +0000 (13:51 +0200)]
libfis: fix wrong shift for lba40 (thanks geoff and charles)

---------- Forwarded message ----------
From: Charles Forsyth <charles.forsyth@gmail.com>
Date: 13 September 2015 at 12:38
Subject: fis bug
To: erik quanstrom <quanstro@quanstro.net>

geoff spotted something similar in sdiahci.c, but it's in libfis as well:

c[Flba24] = lba >> 24;
c[Flba32] = lba >> 32;
c[Flba40] = lba >> 48;

>> 48??  should be >> 40, especially with drive sizes getting up there.

8 years agolibsec: send sigature_algoritms extension for TLS1.2, order ciphers
cinap_lenrek [Sun, 13 Sep 2015 11:39:51 +0000 (13:39 +0200)]
libsec: send sigature_algoritms extension for TLS1.2, order ciphers

TLS1.2 requires the client to send the list of supported
signature and hash algorithm pairs. some servers will simply
reject the client hello otherwise. note that we do not implement
any dh/ecdh param signature verification.

order the cipher list to strogest first. aes128 is actually more
secure than aes256.

8 years agolibsec: handle TLS 1.2 changes in CertificateRequest message
cinap_lenrek [Sat, 12 Sep 2015 11:04:47 +0000 (13:04 +0200)]
libsec: handle TLS 1.2 changes in CertificateRequest message

8 years agocwfs: initialize /env/timezone on boot so dumps are in localtime
cinap_lenrek [Sat, 12 Sep 2015 01:58:46 +0000 (03:58 +0200)]
cwfs: initialize /env/timezone on boot so dumps are in localtime

when /env/timezone file is missing (boot case), copy /adm/timezone/local
to /env/timezone so localtime() can correct timezone offset.

8 years agomkfile.proto: LIB should not be inherited during recursive mk
mischief [Thu, 10 Sep 2015 01:59:40 +0000 (18:59 -0700)]
mkfile.proto: LIB should not be inherited during recursive mk

8 years agohtml2ms: bold table headings, remove wrong tag space handling
cinap_lenrek [Tue, 8 Sep 2015 17:55:55 +0000 (19:55 +0200)]
html2ms: bold table headings, remove wrong tag space handling

8 years agomothra: make text inside <td> bold
cinap_lenrek [Tue, 8 Sep 2015 17:23:23 +0000 (19:23 +0200)]
mothra: make text inside <td> bold

8 years agotar: remove unneccesary { }
cinap_lenrek [Tue, 8 Sep 2015 16:30:38 +0000 (18:30 +0200)]
tar: remove unneccesary { }

8 years agotar: make z flag work, even when no file name was provided (thanks aiju)
cinap_lenrek [Tue, 8 Sep 2015 16:27:48 +0000 (18:27 +0200)]
tar: make z flag work, even when no file name was provided (thanks aiju)

tar used to infer compression type from the filenames extension, but when
no file name is given (stdin/stdout), the -z flag was ignored and no
compression filter applied. this changes tar to assume the default
gzip compression method when z is given and no file name is specified.

8 years agomothra: add missing initializations for plaintext html state
cinap_lenrek [Sun, 6 Sep 2015 22:13:40 +0000 (00:13 +0200)]
mothra: add missing initializations for plaintext html state

8 years agolibthread: get rid of tprivalloc()/tprivfree()/tprivdata() and _workerdata() (thanks...
cinap_lenrek [Sun, 6 Sep 2015 21:25:14 +0000 (23:25 +0200)]
libthread: get rid of tprivalloc()/tprivfree()/tprivdata() and _workerdata() (thanks qrstuv)

these functions where undocumented and unused. especially
tprivfree() was buggy missing a unlock() call. theres not
much point in supporting these functions as theres
threaddata() and procdata().

8 years agolibc: remove privfree(), simplify privalloc()
cinap_lenrek [Sun, 6 Sep 2015 18:43:05 +0000 (20:43 +0200)]
libc: remove privfree(), simplify privalloc()

8 years agodevqspi: fix qunlock error on stat() -> close()
cinap_lenrek [Sat, 5 Sep 2015 11:18:28 +0000 (13:18 +0200)]
devqspi: fix qunlock error on stat() -> close()

8 years agozynq: clean cache unconditionally *before* dma, invalidate cache *after* dma for...
cinap_lenrek [Sat, 5 Sep 2015 08:36:53 +0000 (10:36 +0200)]
zynq: clean cache unconditionally *before* dma, invalidate cache *after* dma for read case

processor might bring data speculatively into the cache,
before the dma completes.

8 years agousbehci: clean cache unconditionally before handing a buffer to the hardware
cinap_lenrek [Sat, 5 Sep 2015 08:14:19 +0000 (10:14 +0200)]
usbehci: clean cache unconditionally before handing a buffer to the hardware

even in the read case, we need to clean the cache
so the cpu will not flush out old changes while
the hardware updates the buffer.