]> git.lizzy.rs Git - plan9front.git/log
plan9front.git
7 years agocfs: fix cfsctl interaction with mount cache (mount -C flag)
cinap_lenrek [Fri, 6 Jan 2017 01:54:26 +0000 (02:54 +0100)]
cfs: fix cfsctl interaction with mount cache (mount -C flag)

the root fileserver is mounted with the mount cache enabled
causing the contents of the cfsctl file to be cached as well
leading to wrong results. so after we generated the new stat
buffer, we increment the qid.vers so the cache starts out
clean.

7 years ago6c, 8c: fix "DI botch" evacuating DI/SI/CX registers to ".save" variables
cinap_lenrek [Mon, 2 Jan 2017 04:49:18 +0000 (05:49 +0100)]
6c, 8c: fix "DI botch" evacuating DI/SI/CX registers to ".save" variables

7 years agovgaigfx: enable softscreen by default
cinap_lenrek [Thu, 29 Dec 2016 02:46:08 +0000 (03:46 +0100)]
vgaigfx: enable softscreen by default

given that the igfx driver doesnt provide any acceleration functions
and drawing is usually faster with double buffering as it eleminates
reads over the pci bus, enable softscreen by default.

7 years agosamterm: avoid flushimage when theres nothing new to flush
cinap_lenrek [Wed, 28 Dec 2016 22:42:56 +0000 (23:42 +0100)]
samterm: avoid flushimage when theres nothing new to flush

7 years agolibmp: avoid temporary buffer allocation in mprand()
cinap_lenrek [Wed, 28 Dec 2016 17:19:52 +0000 (18:19 +0100)]
libmp: avoid temporary buffer allocation in mprand()

7 years agofplot: fix zoom egetrect check (thanks qu7uux)
cinap_lenrek [Wed, 28 Dec 2016 01:09:33 +0000 (02:09 +0100)]
fplot: fix zoom egetrect check (thanks qu7uux)

7 years agofplot: parse negative numbers in argument to -r option (thanks qu7uux)
cinap_lenrek [Wed, 28 Dec 2016 01:08:45 +0000 (02:08 +0100)]
fplot: parse negative numbers in argument to -r option (thanks qu7uux)

7 years agolibsec: replace des based X9.17 genrandom() with chacha random number generator
cinap_lenrek [Wed, 28 Dec 2016 01:02:00 +0000 (02:02 +0100)]
libsec: replace des based X9.17 genrandom() with chacha random number generator

7 years agokeyfs: print error message when reading /adm/keys fails
cinap_lenrek [Mon, 26 Dec 2016 18:48:38 +0000 (19:48 +0100)]
keyfs: print error message when reading /adm/keys fails

7 years agocdproto: explicitely create /$objtype/bin subdirectories for all archs
cinap_lenrek [Sun, 25 Dec 2016 01:13:17 +0000 (02:13 +0100)]
cdproto: explicitely create /$objtype/bin subdirectories for all archs

7 years agolibavl: fix manpage example, minor improvement to code
spew [Sat, 24 Dec 2016 23:58:31 +0000 (17:58 -0600)]
libavl: fix manpage example, minor improvement to code

7 years agoavl: fix man page example
spew [Fri, 23 Dec 2016 00:44:45 +0000 (18:44 -0600)]
avl: fix man page example

7 years agoalv(2): new avl implementation
spew [Thu, 22 Dec 2016 22:47:41 +0000 (16:47 -0600)]
alv(2): new avl implementation

7 years agoauth/as: simplify further
cinap_lenrek [Thu, 22 Dec 2016 20:55:56 +0000 (21:55 +0100)]
auth/as: simplify further

7 years agoauth/as, auth/none, auth/newns: consistent handling of command arguments, cleanup
cinap_lenrek [Thu, 22 Dec 2016 20:39:59 +0000 (21:39 +0100)]
auth/as, auth/none, auth/newns: consistent handling of command arguments, cleanup

7 years agomerge
cinap_lenrek [Thu, 22 Dec 2016 00:59:00 +0000 (01:59 +0100)]
merge

7 years agolibauth: don't attempt to mount when opening mount srv file fails in nsop()
cinap_lenrek [Thu, 22 Dec 2016 00:56:19 +0000 (01:56 +0100)]
libauth: don't attempt to mount when opening mount srv file fails in nsop()

making newnsdebug error messages more usefull...

7 years agompxor: sign should be 1/-1, not 0/-1
aiju [Tue, 20 Dec 2016 17:33:58 +0000 (18:33 +0100)]
mpxor: sign should be 1/-1, not 0/-1

7 years agorcpu: avoid filedescriptor conflict with <{} (thanks mycroftiv)
cinap_lenrek [Mon, 19 Dec 2016 18:05:35 +0000 (19:05 +0100)]
rcpu: avoid filedescriptor conflict with <{} (thanks mycroftiv)

the rcpu client dup's fd 0,1,2 to fd 10,11,12 which
can accidentally override the pipe file descriptor
allocated by the <{} operator. to avoid this problem,
we generate the remote script as an /env file in a
separate step now.

7 years agoawk: improve random number generation
cinap_lenrek [Sun, 18 Dec 2016 17:00:45 +0000 (18:00 +0100)]
awk: improve random number generation

don't use rand() and scale it to 0..1, instead call
native frand() which produces uniform random number.

instead of seeding the rng with time(0), use truerand().

7 years agopc: modify cpu0 page tables in patwc() instead of current cpu ones
cinap_lenrek [Sat, 17 Dec 2016 18:47:35 +0000 (19:47 +0100)]
pc: modify cpu0 page tables in patwc() instead of current cpu ones

on 386 kernel, each processor has its own pdb where the primary
pdb for kernel mappings is on cpu0 and other cpu's lazily pull
pdb entries from cpu0 when they fault in vmapsync().

so we have to edit the table tables in the pdb of cpu0 and not
the current processor.

7 years agopat write combinding support for 386 kernel, honor cpuid bits
cinap_lenrek [Sat, 17 Dec 2016 15:35:26 +0000 (16:35 +0100)]
pat write combinding support for 386 kernel, honor cpuid bits

7 years agopc64: implement simple write combining for framebuffers with the PAT
cinap_lenrek [Thu, 15 Dec 2016 22:27:01 +0000 (23:27 +0100)]
pc64: implement simple write combining for framebuffers with the PAT

on some modern machines like the x250, the bios arranges the mtrr's
and the framebuffer membar in a way that doesnt allow us to mark
the framebuffer pages as write combining, leading to slow graphics.

since the pentium III, the processor interprets the page table bit
combinations of the WT, CD and bit7 bits as an index into the
page attribute table (PAT).

to not change the semantics of the WT and CD bits, we preserve
the bit patterns 0-3 and use the last entry 7 for write combining.
(done in mmuinit() for each core).

the new patwc() function takes virtual address range and changes
the page table marking the range as write combining. no attempt
is made on invalidating tlb's. doesnt matter in our case as the
following mtrr() call in screen.c does it for us.

7 years agoacid: fix y.tab.h dependency for proc.$O
cinap_lenrek [Thu, 15 Dec 2016 21:08:57 +0000 (22:08 +0100)]
acid: fix y.tab.h dependency for proc.$O

7 years agoxen: fix build by adding missing rdrandbuf() function
cinap_lenrek [Wed, 14 Dec 2016 19:02:07 +0000 (20:02 +0100)]
xen: fix build by adding missing rdrandbuf() function

7 years agovncs: update devmouse code
cinap_lenrek [Sun, 11 Dec 2016 15:53:51 +0000 (16:53 +0100)]
vncs: update devmouse code

7 years agovnc: don't prompt for password on auth_respond() failure
cinap_lenrek [Sun, 11 Dec 2016 15:52:38 +0000 (16:52 +0100)]
vnc: don't prompt for password on auth_respond() failure

7 years agodevmouse: remove unused static map[] array
cinap_lenrek [Sun, 11 Dec 2016 15:02:30 +0000 (16:02 +0100)]
devmouse: remove unused static map[] array

7 years agodevmouse: change msec argument of *mousetrack() to ulong
cinap_lenrek [Sat, 10 Dec 2016 15:12:18 +0000 (16:12 +0100)]
devmouse: change msec argument of *mousetrack() to ulong

7 years agoaudioac97: support for ICH4-ICH7 based cards with memory mapped registers (thanks...
cinap_lenrek [Mon, 5 Dec 2016 17:32:58 +0000 (18:32 +0100)]
audioac97: support for ICH4-ICH7 based cards with memory mapped registers (thanks echoline)

7 years agotcp17019: make service proto and netdir arguments optional (for aux/listen1)
cinap_lenrek [Fri, 2 Dec 2016 23:24:19 +0000 (00:24 +0100)]
tcp17019: make service proto and netdir arguments optional (for aux/listen1)

this allows /rc/bin/service/tcp17019 to be called from
aux/listen1 without arguments like:

aux/listen1 tcp!*!rcpu /rc/bin/service/tcp17019

7 years agodevmouse: various bugfixes, simplify
cinap_lenrek [Tue, 29 Nov 2016 20:11:48 +0000 (21:11 +0100)]
devmouse: various bugfixes, simplify

the assumption of only one producer ((abs)moustratrack()) is not true
for external mouse events from /dev/mousein, so protect the mouse state
and queue with ilock().

get rid of mousecreate(), just use devcreate().

reset cursor when all instances of /dev/mouse and /dev/cursor got closed,
instead of also considering /dev/mousectl. the reason is that kbdfs keeps
the mousectl file open. so exiting a program that has the cursor changed
will properly reset the cursor to arrow.

don't access user buffer while holding cursor spinlock! the memory access
can fault. theres also no lock needed there, we'r just copying *from* the
cursor memory.

fix use of strtol(), p will always be set, check for end of string.

keep pointer coordinates onscreen (off by one).

make lastms() function to get the last millisecond delta of last
call for resynchronization.

fix msg[3] buffer overflow in m5mouseputc().

get rid of mouseshifted logic, it is not used.

7 years agoomap: cleanup mouse.c, just a dummy for mousectl()
cinap_lenrek [Tue, 29 Nov 2016 19:54:17 +0000 (20:54 +0100)]
omap: cleanup mouse.c, just a dummy for mousectl()

7 years agoomap: fix format string warning %d for long
cinap_lenrek [Tue, 29 Nov 2016 19:51:08 +0000 (20:51 +0100)]
omap: fix format string warning %d for long

7 years agostdio: fix sclose() buffer overrun when terminating string, realloc() error handling...
cinap_lenrek [Sun, 27 Nov 2016 20:20:27 +0000 (21:20 +0100)]
stdio: fix sclose() buffer overrun when terminating string, realloc() error handling (thanks porlock)

theres a bug is in sclose() where it doesnt check if wp is beyond
the buffer. also wp was not updated after realloc().

bug was reported by porlock on 9fans:

Plan 9's implementation of the standard C functions snprintf and
vsnprintf have a buffer overrun bug.

If the buffer length equals the output length (without the terminating
null), then one too many characters is written to the buffer.

For example,
              snprintf(buf, 4, "ABCD");

will write 5 characters to buf.

7 years agomerge
cinap_lenrek [Sat, 19 Nov 2016 15:47:21 +0000 (16:47 +0100)]
merge

7 years agolink loopbackmedium and netdevmedium in bcm/pif, sgi/indy and zynq kernels
cinap_lenrek [Sat, 19 Nov 2016 15:45:18 +0000 (16:45 +0100)]
link loopbackmedium and netdevmedium in bcm/pif, sgi/indy and zynq kernels

7 years agolibavl, libregexp: put debug functions back
ftrvxmtrx [Thu, 17 Nov 2016 22:07:54 +0000 (23:07 +0100)]
libavl, libregexp: put debug functions back

7 years ago/sys/src/lib*: clean up
ftrvxmtrx [Thu, 17 Nov 2016 21:58:16 +0000 (22:58 +0100)]
/sys/src/lib*: clean up

7 years agolibmemdraw: cleanup fillpoly(), remove unused fillcolor hack
cinap_lenrek [Thu, 17 Nov 2016 19:10:07 +0000 (20:10 +0100)]
libmemdraw: cleanup fillpoly(), remove unused fillcolor hack

7 years agolibsec: remove unused get32() function
cinap_lenrek [Thu, 17 Nov 2016 18:54:13 +0000 (19:54 +0100)]
libsec: remove unused get32() function

7 years agolibmemdraw: remove unused static drawbuf variables and ptrfn() declaration
cinap_lenrek [Thu, 17 Nov 2016 18:42:12 +0000 (19:42 +0100)]
libmemdraw: remove unused static drawbuf variables and ptrfn() declaration

7 years agodevtls: remove unused get32() function
cinap_lenrek [Thu, 17 Nov 2016 17:14:06 +0000 (18:14 +0100)]
devtls: remove unused get32() function

7 years agocmd: remove a bit of unused stuff
ftrvxmtrx [Thu, 17 Nov 2016 01:59:40 +0000 (02:59 +0100)]
cmd: remove a bit of unused stuff

7 years agotar: remove unused variable
ftrvxmtrx [Thu, 17 Nov 2016 01:14:59 +0000 (02:14 +0100)]
tar: remove unused variable

7 years agoclock: remove unused variable
ftrvxmtrx [Thu, 17 Nov 2016 01:12:49 +0000 (02:12 +0100)]
clock: remove unused variable

7 years agopage: remove unused variables
ftrvxmtrx [Thu, 17 Nov 2016 01:11:35 +0000 (02:11 +0100)]
page: remove unused variables

7 years agoramfs: remove unused variable
ftrvxmtrx [Thu, 17 Nov 2016 01:09:57 +0000 (02:09 +0100)]
ramfs: remove unused variable

7 years agoaan: didn't ask about sendcommand
ftrvxmtrx [Thu, 17 Nov 2016 01:07:52 +0000 (02:07 +0100)]
aan: didn't ask about sendcommand

7 years agolibsec: remove unused aes_setupDec
ftrvxmtrx [Thu, 17 Nov 2016 01:02:32 +0000 (02:02 +0100)]
libsec: remove unused aes_setupDec

7 years agopc64: check if vmap() range fits in VMAPLEN window, remove unneeded vmapsync(), renam...
cinap_lenrek [Thu, 17 Nov 2016 00:28:11 +0000 (01:28 +0100)]
pc64: check if vmap() range fits in VMAPLEN window, remove unneeded vmapsync(), rename fault386() to faultamd64()

7 years agoip/tcp: never raise the mss over the link mtu < 1280 for v6
cinap_lenrek [Tue, 15 Nov 2016 23:54:04 +0000 (00:54 +0100)]
ip/tcp: never raise the mss over the link mtu < 1280 for v6

v6 mandates minimum mtu of 1280, tho someone *could* setup
an interface with a lower mtu or set it lower for testing.

7 years agoip: get rid of update_mtucache() and restrict_mtu() prototypes
cinap_lenrek [Tue, 15 Nov 2016 21:13:08 +0000 (22:13 +0100)]
ip: get rid of update_mtucache() and restrict_mtu() prototypes

7 years agoip/pktmedium: no mintu, no maclen... thi is ip packets
cinap_lenrek [Tue, 15 Nov 2016 21:11:47 +0000 (22:11 +0100)]
ip/pktmedium: no mintu, no maclen... thi is ip packets

7 years agoip/pktmedium: fix wrong hsize, theres no ethernet header on packet media
cinap_lenrek [Tue, 15 Nov 2016 20:54:03 +0000 (21:54 +0100)]
ip/pktmedium: fix wrong hsize, theres no ethernet header on packet media

packet media is just raw ip packets, so theres no link-level
header there. was probably copy-pasted from ethermedium...

7 years agoip/tcp: only calculae mss from interface mtu when directly reachable for v6
cinap_lenrek [Tue, 15 Nov 2016 19:28:45 +0000 (20:28 +0100)]
ip/tcp: only calculae mss from interface mtu when directly reachable for v6

we currently do not implement path mtu discovery so for
destinations that are not directly reachable assume the
minimum mtu of 1280 bytes.

7 years agonusb/ether: support for "bridge" ctl message
cinap_lenrek [Sat, 12 Nov 2016 22:11:03 +0000 (23:11 +0100)]
nusb/ether: support for "bridge" ctl message

7 years agokernel/qio: make readblist() offset of type ulong as the rest
cinap_lenrek [Sat, 12 Nov 2016 16:41:58 +0000 (17:41 +0100)]
kernel/qio: make readblist() offset of type ulong as the rest

7 years agokernel/qio: get rid of unused qcopycnt debug variable
cinap_lenrek [Sat, 12 Nov 2016 15:05:00 +0000 (16:05 +0100)]
kernel/qio: get rid of unused qcopycnt debug variable

7 years agokernel/qio: implement concatblock() with pullupblock()
cinap_lenrek [Wed, 9 Nov 2016 18:44:38 +0000 (19:44 +0100)]
kernel/qio: implement concatblock() with pullupblock()

7 years agokernel/ip: fix typo (rfc -> ifc)
cinap_lenrek [Tue, 8 Nov 2016 21:33:48 +0000 (22:33 +0100)]
kernel/ip: fix typo (rfc -> ifc)

7 years agogre: check nil for pullupblock()
cinap_lenrek [Tue, 8 Nov 2016 21:33:19 +0000 (22:33 +0100)]
gre: check nil for pullupblock()

7 years agodevbridge: simplify etherwrite() as we dont deal with block lists
cinap_lenrek [Tue, 8 Nov 2016 20:06:06 +0000 (21:06 +0100)]
devbridge: simplify etherwrite() as we dont deal with block lists

7 years agokernel/ip: remove nil checks for allocb() and padblock()
cinap_lenrek [Tue, 8 Nov 2016 20:05:01 +0000 (21:05 +0100)]
kernel/ip: remove nil checks for allocb() and padblock()

7 years agokernel/qio: maintain Block.next pointer in padblock()
cinap_lenrek [Tue, 8 Nov 2016 20:03:08 +0000 (21:03 +0100)]
kernel/qio: maintain Block.next pointer in padblock()

7 years agosrv(4): add SOURCE for srvtls
cinap_lenrek [Tue, 8 Nov 2016 01:10:47 +0000 (02:10 +0100)]
srv(4): add SOURCE for srvtls

7 years agokernel/qio: fix comments, fix qiwrite() on close queue, remove debug setmalloctag...
cinap_lenrek [Mon, 7 Nov 2016 23:38:07 +0000 (00:38 +0100)]
kernel/qio: fix comments, fix qiwrite() on close queue, remove debug setmalloctag() call in qwrite()

7 years agodevcons: simplify putstrn0()
cinap_lenrek [Mon, 7 Nov 2016 23:34:59 +0000 (00:34 +0100)]
devcons: simplify putstrn0()

7 years agodevbridge: various bugfixes and improvements from charles forsyth
cinap_lenrek [Mon, 7 Nov 2016 21:43:37 +0000 (22:43 +0100)]
devbridge: various bugfixes and improvements from charles forsyth

7 years agokernel/qio: big cleanup of qio functions
cinap_lenrek [Mon, 7 Nov 2016 21:20:10 +0000 (22:20 +0100)]
kernel/qio: big cleanup of qio functions

remove bl2mem(), it is broken. a fault while copying to memory
yields a partially freed block list. it can be simply replaced
by readblist() and freeblist(), which we also use for qcopy()
now.

remove mem2bl(), and handle putting back remainer from a short
read internally (splitblock()) avoiding the releasing and re-
acquiering of the ilock.

always attempt to free blocks outside of the ilock.

have qaddlist() return the number of bytes enqueued, which
avoids walking the block list twice.

7 years agodevloopback: simplify loopoput()
cinap_lenrek [Mon, 7 Nov 2016 21:08:21 +0000 (22:08 +0100)]
devloopback: simplify loopoput()

remove unneeded waserror() block, loopoput is alled from
loopbackbwrite only so we will always get called with a
*single* block, so the concatblock() is not needed.

7 years agoip: always pass a single block to Medium.bwrite(), avoid concatblock() calls in Dev...
cinap_lenrek [Mon, 7 Nov 2016 21:05:29 +0000 (22:05 +0100)]
ip: always pass a single block to Medium.bwrite(), avoid concatblock() calls in Dev.bwrite()

the convention for Dev.bwrite() is that it accepts a *single* block,
and not a block chain. so we never have concatblock here.

to keep stuff consistent, we also guarantee thet Medium.bwrite()
will get a *single* block passed as well, as the callers are
few in number.

7 years agoip/nullmedium: free passed block in nullbwrite()
cinap_lenrek [Mon, 7 Nov 2016 20:40:12 +0000 (21:40 +0100)]
ip/nullmedium: free passed block in nullbwrite()

7 years agoip/tcp: remove useless nil checks for padblock() and allocb() return value
cinap_lenrek [Mon, 7 Nov 2016 20:39:28 +0000 (21:39 +0100)]
ip/tcp: remove useless nil checks for padblock() and allocb() return value

7 years agokernel: fix missing ; in panic() call
cinap_lenrek [Sat, 5 Nov 2016 19:08:20 +0000 (20:08 +0100)]
kernel: fix missing ; in panic() call

7 years agokernel: avoid padblock copying for devtls/devssl/esp, cleanup debugging
cinap_lenrek [Sat, 5 Nov 2016 19:05:40 +0000 (20:05 +0100)]
kernel: avoid padblock copying for devtls/devssl/esp, cleanup debugging

to avoid copying in padblock() when adding cryptographics macs to a block
in devtls/devssl/esp we reserve 16 extra bytes to the allocation.

remove qio ixsummary() function and add acid function qiostats() to
/sys/lib/acid/kernel

simplify iallocb(), remove iallocsummary() statitics.

7 years agoarchacpi: fix format string warning on amd64
cinap_lenrek [Sat, 5 Nov 2016 17:31:50 +0000 (18:31 +0100)]
archacpi: fix format string warning on amd64

7 years agodevmnt: avoid memory copies of I/O rpc buffer by using bwrite()
cinap_lenrek [Sat, 5 Nov 2016 17:26:12 +0000 (18:26 +0100)]
devmnt: avoid memory copies of I/O rpc buffer by using bwrite()

given that devmnt will almost always write into a pipe
or a network connection, which supports te bwrite routine,
we can avoid the memory copy that would have been done by
devbwrite(). this also means the i/o buffer for writes
will get freed sooner without having to wait for the 9p
rpc to get a response, saving memory.

theres one case where we have to keep the rpc arround and
that is when we write to a cached file, as we want to update
the cache with the data that was written, but the user buffer
cannot be trusted to stay the same during the rpc.

7 years agolibc: move calloc() into its own compilation unit
cinap_lenrek [Sat, 5 Nov 2016 17:00:10 +0000 (18:00 +0100)]
libc: move calloc() into its own compilation unit

move calloc() in its own compilation unit to avoid
code duplication. also, calloc() is used rarely in
plan9 programs.

7 years agowinwatch: put previous label contents in edit buffer for relabling
cinap_lenrek [Tue, 1 Nov 2016 08:50:25 +0000 (09:50 +0100)]
winwatch: put previous label contents in edit buffer for relabling

7 years ago8c: fix double compiling FNX complex lvalue in cgen64()
cinap_lenrek [Sun, 30 Oct 2016 22:30:13 +0000 (23:30 +0100)]
8c: fix double compiling FNX complex lvalue in cgen64()

sugen() calls cgen64() speculatively so that when cgen64() returns
zero, it will fall back and compile 64-bit copy.

the bug was that cgen64() compiled the left hand side and then recursively
called cgen64() again, which didnt handle the memory copy so it returned
zero and sugen() would compile the left hand side again resulting in two
function calls being emited.

some code that reproduced the issue:

#include <u.h>
#include <libc.h>

typedef struct
{
char x[10];
vlong a;
} X;

X a;
X *f(void) { return &a; }

void
main(int argc, char *argv[])
{
f()->a = a.a;
}

producing:

TEXT f+0(SB),0,$0
MOVL $a+0(SB),AX
RET ,
RET ,
TEXT main+0(SB),0,$0
CALL ,f+0(SB)
CALL ,f+0(SB) <- bug
MOVL AX,CX
LEAL a+12(SB),DX
MOVL (DX),AX
MOVL AX,12(CX)
MOVL 4(DX),AX
MOVL AX,16(CX)
RET ,
GLOBL a+0(SB),$20
END ,

7 years agoape/libsec: add secp384r1 curve parameters for tls
cinap_lenrek [Sun, 30 Oct 2016 19:32:03 +0000 (20:32 +0100)]
ape/libsec: add secp384r1 curve parameters for tls

7 years agomerge
cinap_lenrek [Sun, 30 Oct 2016 01:17:12 +0000 (02:17 +0100)]
merge

7 years agolibsec: add secp384r1 curve parameters for tls
cinap_lenrek [Sun, 30 Oct 2016 01:15:40 +0000 (02:15 +0100)]
libsec: add secp384r1 curve parameters for tls

7 years agowinwatch: no more 'l' key labeling
spew [Wed, 26 Oct 2016 19:01:13 +0000 (14:01 -0500)]
winwatch: no more 'l' key labeling

7 years agowinwatch: middle mouse button click also prompts for label
spew [Wed, 26 Oct 2016 03:53:38 +0000 (22:53 -0500)]
winwatch: middle mouse button click also prompts for label

7 years agostats: alternatively read battery and temp from aux/acpi
ftrvxmtrx [Mon, 24 Oct 2016 21:51:16 +0000 (23:51 +0200)]
stats: alternatively read battery and temp from aux/acpi

7 years agoaux/acpi: make it stats(8)-friendly
ftrvxmtrx [Mon, 24 Oct 2016 21:50:04 +0000 (23:50 +0200)]
aux/acpi: make it stats(8)-friendly

7 years agofix bugs and cleanup cryptsetup code
cinap_lenrek [Mon, 24 Oct 2016 18:56:11 +0000 (20:56 +0200)]
fix bugs and cleanup cryptsetup code

devfs:

- fix memory leak in devfs leaking the aes key
- allocate aes-xts cipher state in secure memory
- actually check if the hexkey got fully parsed

cryptsetup:

- get rid of stupid "type YES" prompt
- use genrandom() to generate salts and keys
- rewrite cryptsetup to use common pbkdf2 and readcons routines
- fix alot of error handling and simplify the code
- move cryptsetup command to disk/cryptsetup
- update cryptsetup(8) manual page

7 years agoauth/factotum: bound the number of srv processes to 16, error the 9p rpc when it...
cinap_lenrek [Sun, 23 Oct 2016 15:18:35 +0000 (17:18 +0200)]
auth/factotum: bound the number of srv processes to 16, error the 9p rpc when it is over limit

7 years agolib9p: limit the number of srv processes kept arround 8
cinap_lenrek [Sun, 23 Oct 2016 15:16:22 +0000 (17:16 +0200)]
lib9p: limit the number of srv processes kept arround 8

7 years agopc/archacpi: don't do acpi initialization in the kernel, provide generic acpimem...
cinap_lenrek [Sun, 23 Oct 2016 02:09:27 +0000 (04:09 +0200)]
pc/archacpi: don't do acpi initialization in the kernel, provide generic acpimem routines

get rid of _INI and _REG method calls, this is not full acpi environment
anyway and all we really want todo at kernel boot time is figuring out
the interrupt routing. aux/acpi can try to enable more stuff if it needs
to later when battery status desired.

dont snoop memory space regions in amlmapio(), this is just wrong as
amlmapio() is *lazily* mapping regions as they are accessed, so the
range table would never be really complete. instead, we provide generic
access to the physical address space, excluding kernel and user memory
with acpimem file.

7 years agoip: simplify code as packblock() and concatblock() will never error
cinap_lenrek [Sat, 22 Oct 2016 22:31:42 +0000 (00:31 +0200)]
ip: simplify code as packblock() and concatblock() will never error

7 years agodevip: simplify ipbwrite() by using retun value of qbwrite()
cinap_lenrek [Sat, 22 Oct 2016 22:29:41 +0000 (00:29 +0200)]
devip: simplify ipbwrite() by using retun value of qbwrite()

7 years agoip/icmp: only reply to echo request when directed to us and source is unicast
cinap_lenrek [Sat, 22 Oct 2016 22:25:17 +0000 (00:25 +0200)]
ip/icmp: only reply to echo request when directed to us and source is unicast

7 years agogif: just read and decode the first image when -9 or -c flag is specified
cinap_lenrek [Fri, 21 Oct 2016 22:35:17 +0000 (00:35 +0200)]
gif: just read and decode the first image when -9 or -c flag is specified

7 years agolibdraw: avoid dropping queued button change mouse events in emouse()
cinap_lenrek [Fri, 21 Oct 2016 22:04:21 +0000 (00:04 +0200)]
libdraw: avoid dropping queued button change mouse events in emouse()

7 years agomothra: avoid flushimage() calls, event() will implicitely flush
cinap_lenrek [Fri, 21 Oct 2016 22:02:23 +0000 (00:02 +0200)]
mothra: avoid flushimage() calls, event() will implicitely flush

7 years agompc: remove unused yylval type
spew [Thu, 20 Oct 2016 03:15:51 +0000 (22:15 -0500)]
mpc: remove unused yylval type

7 years agoaux/acpi: batteries and CPU temp reading so far
ftrvxmtrx [Tue, 18 Oct 2016 21:05:09 +0000 (23:05 +0200)]
aux/acpi: batteries and CPU temp reading so far