martynas [Sat, 18 Jan 2014 05:54:51 +0000 (05:54 +0000)]
Remove -Wbounded: it is now the compiler default.
dlg [Sat, 18 Jan 2014 04:24:11 +0000 (04:24 +0000)]
take advantage of pool_setipl and PR_ZERO to avoid doing work the
subsystem provides for us now when dealing with the xfer pool.
ok krw@
dlg [Sat, 18 Jan 2014 04:11:07 +0000 (04:11 +0000)]
if the controller is too busy to do a POLLed command, return XS_BUSY
instead of XS_NO_CCB.
ok krw@
jsing [Sat, 18 Jan 2014 03:07:05 +0000 (03:07 +0000)]
Make installboot(8) easier to use - copy the second stage boot loader to
the default location used by the given architecture. This eliminates the
need to copy it over manually prior to running installboot.
jsing [Sat, 18 Jan 2014 02:47:27 +0000 (02:47 +0000)]
Add a -r flag that allows for the mount point of the root filesystem to be
specified. This is primarily for use by the installer and defaults to /.
jsing [Sat, 18 Jan 2014 02:45:38 +0000 (02:45 +0000)]
Build installboot on all architectures.
dlg [Sat, 18 Jan 2014 02:42:30 +0000 (02:42 +0000)]
rename scsi_ioh_runqueue to scsi_iopool_run, and make it available
outside scsi_base.c.
this will allow adapters to restrict access to iopool resources
based on some state, and then kick the pending requests on the pool
when the state comes good again.
ive been avoiding this for a long time, but it is the least worst
way to deal with some uses of XS_NO_CCB.
discussion with kettenis@ helped me decide this was right.
jsing [Sat, 18 Jan 2014 02:31:18 +0000 (02:31 +0000)]
Also make installboot build on macppc.
deraadt [Sat, 18 Jan 2014 01:58:33 +0000 (01:58 +0000)]
Use arc4random_uniform to increase the slightly difficulty of winning.
Kidding, the fix is to stop using the scheduler's PRNG.
ok kettenis
jsing [Sat, 18 Jan 2014 01:49:30 +0000 (01:49 +0000)]
Add missing include.
jsing [Sat, 18 Jan 2014 01:44:56 +0000 (01:44 +0000)]
Make installboot(8) build on all architectures, although currently with
non-functional stubs for alpha, armish, armv7, aviion, loongson, luna88k,
mvme68k, mvme88k, octeon, sgi, socppc, sparc and zaurus.
krw [Sat, 18 Jan 2014 01:41:28 +0000 (01:41 +0000)]
If ugflag (indicating a character from the file being parsed has been
unget()'ed) is set, then properly set the lexchar position back one in
get_token() so more error messages put the '^' in the proper spot.
krw [Sat, 18 Jan 2014 01:19:01 +0000 (01:19 +0000)]
Remove superfluous assignment to local variable 'val' just before
exiting.
espie [Sat, 18 Jan 2014 01:10:36 +0000 (01:10 +0000)]
involuntary sneak preview, revert
espie [Sat, 18 Jan 2014 01:09:30 +0000 (01:09 +0000)]
...and sort SHA256 properly
krw [Sat, 18 Jan 2014 01:03:56 +0000 (01:03 +0000)]
Make parse_warn() messages consistantly use 'expecting' rather
than occasionally 'expected'. End all with a '.'.
espie [Sat, 18 Jan 2014 01:02:25 +0000 (01:02 +0000)]
caching a bit more tricky... problem noticed by naddy@
krw [Sat, 18 Jan 2014 00:51:55 +0000 (00:51 +0000)]
Never silently consume the following statement when unexpectedly
encountering a ';'. I.e. when checking the token type, 'skip_to_semi()'
after 'parse_warn()' only when the parsed token wasn't a ';'.
naddy [Sat, 18 Jan 2014 00:18:06 +0000 (00:18 +0000)]
correct crypto lingo; ok espie@
dlg [Fri, 17 Jan 2014 23:23:58 +0000 (23:23 +0000)]
if the hardware is too busy or its queues are too full, return XS_BUSY,
not XS_NO_CCB.
discussed with krw@ who agrees
tobias [Fri, 17 Jan 2014 23:13:49 +0000 (23:13 +0000)]
Validate file size before loading content into memory.
ok millert@
tobias [Fri, 17 Jan 2014 23:11:42 +0000 (23:11 +0000)]
Do not follow symlinks for -l and -L arguments.
ok krw@
espie [Fri, 17 Jan 2014 22:59:44 +0000 (22:59 +0000)]
signing got simpler
dlg [Fri, 17 Jan 2014 22:52:32 +0000 (22:52 +0000)]
a polled command timing out isnt a software resource shortage, its
a driver stuffup.
aac is XS_NO_CCB clean now.
dlg [Fri, 17 Jan 2014 22:51:10 +0000 (22:51 +0000)]
remove a XS_NO_CCB case by moving aac to iopools.
eyes and ok by krw@ jmatthew@
krw [Fri, 17 Jan 2014 22:48:10 +0000 (22:48 +0000)]
Make parse_X return -1 when it encounters a parsing error. Enables
recognition of zero length value vs parsing error. This lets us avoid
the erroneous consumption of the semicolon ending the 'X' value, and
the resulting accidental consumption of the following line in the file
being parsed.
dlg [Fri, 17 Jan 2014 22:20:32 +0000 (22:20 +0000)]
replace aacminphys with scsi_minphys. both are just wrappers around
minphys.
dlg [Fri, 17 Jan 2014 22:18:27 +0000 (22:18 +0000)]
aacvar advertises a public api which is only used inside aac.c. make it
"private" so i dont feel bad about changing it.
tobias [Fri, 17 Jan 2014 21:42:47 +0000 (21:42 +0000)]
Close file descriptor before next try getting a good one.
ok millert@
tobias [Fri, 17 Jan 2014 20:51:54 +0000 (20:51 +0000)]
Fix off-by-one on specially crafted /etc/shells file.
ok gilles@, millert@
schwarze [Fri, 17 Jan 2014 19:22:45 +0000 (19:22 +0000)]
Sort the macro keys by their real-world frequency to reduce the average
mask size. No functional change.
This shrinks the standard /usr/share/man database by 7%, now at 10.3x
the size of whatis.db, and with -Q even by 11%, now at 3.0x of whatis.db.
Now i'm out of ideas to easily shrink the size of the database.
Optimization found somewhere above the West Australian desert.
Committing from Melbourne, Victoria.
schwarze [Fri, 17 Jan 2014 19:16:42 +0000 (19:16 +0000)]
Drop the AUTOINCREMENT PRIMARY KEYs from the mlinks and keys tables.
They are completely unused, and i cannot imagine what they *could*
ever be used for; but apparently, they are expensive to generate.
Standard DB build time goes down by 10%, now at 1.9x of makewhatis.
Standard DB size goes down by 4%, now at 11x of makewhatis.
DB build time with -Q goes down by 15%, now at 0.28x of makewhatis.
DB size with -Q goes down by 3%, now at 3.35x of makewhatis.
Optimization found somewhere above the Southern Balochistan desert,
near the Iran-Pakistan border. Committing from Melbourne, Victoria.
schwarze [Fri, 17 Jan 2014 19:07:12 +0000 (19:07 +0000)]
Despite some experimenting, i'm unable to find any relevant effect of
creating an index for the keys table on apropos(1) search times;
apparently, adding that index was premature optimization in the first
place; so, stop adding that index.
Its root gone, the following evil is reduced (/usr/share/man on my notebook)
- DB build time with -Q goes down by 15%, now at 1/3 of makewhatis
- DB size with -Q goes down by 35%, now at 3.5x of makewhatis
- full DB build time goes down by 12%, now at 2.1x of makewhatis
- full DB size goes down by 42%, now at 11.5x of makewhatis
Optimization found somewhere above the Dasht-e Kavir, the Great Salt
Desert, between Tehran (Iran) and Herat (Afghanistan). Committing
from Melbourne, Australia.
okan [Fri, 17 Jan 2014 18:42:30 +0000 (18:42 +0000)]
ctype and other sign extension fixes.
with deraadt and millert, ok millert
espie [Fri, 17 Jan 2014 15:54:06 +0000 (15:54 +0000)]
do check_digest per plist.
show UNSIGNED packages unconditionally.
espie [Fri, 17 Jan 2014 15:46:16 +0000 (15:46 +0000)]
make ArcCheck less confusing, don't archive stuff that WON'T survive
extraction anyways (those metadatas ARE in the packing-list anyways).
espie [Fri, 17 Jan 2014 15:39:53 +0000 (15:39 +0000)]
"recognize" extended header thingies (we don't handle them, we just
spew more useful error messages)
espie [Fri, 17 Jan 2014 13:41:47 +0000 (13:41 +0000)]
use internal gunzip, now that the fh leak is closed.
espie [Fri, 17 Jan 2014 13:15:43 +0000 (13:15 +0000)]
oops, keeping _current around means I still have a live fh after deref.
So remove it too. Accordingly, the code that resyncs archives on reopen
needs us tracking the name itself.
espie [Fri, 17 Jan 2014 11:09:36 +0000 (11:09 +0000)]
tweak documentation to match what's going on
espie [Fri, 17 Jan 2014 10:59:18 +0000 (10:59 +0000)]
as requested by theo, invoke signify in the EXAMPLE order.
espie [Fri, 17 Jan 2014 10:55:01 +0000 (10:55 +0000)]
default signer less picky, just select by function, e.g., *pkg or *fw.
espie [Fri, 17 Jan 2014 10:54:14 +0000 (10:54 +0000)]
tweak the interface to generating signatures yet again.
- assume key names match, deduce signer from sec key.
e.g., -s signify -s 55pkg.sec
will set signer to 55pkg and look for a pubkey named 55pkg.pub,
either besides 55pkg.sec or in /etc/signify.
- verify there's no mismatch, if possible, by verifying the first package
signed.
- also build a SHA256 on the fly while signing.
dtucker [Fri, 17 Jan 2014 06:23:24 +0000 (06:23 +0000)]
fix log message statvfs. ok djm
dtucker [Fri, 17 Jan 2014 05:26:41 +0000 (05:26 +0000)]
remove unused includes. ok djm@
lteo [Fri, 17 Jan 2014 03:38:12 +0000 (03:38 +0000)]
Add a new example showing how to verify bsd.rd with signify and the new
sha256 -C option.
suggested by deraadt@
help/OK jmc@ tedu@
halex [Fri, 17 Jan 2014 01:15:27 +0000 (01:15 +0000)]
add signature checking and make checksum procedure more robust by
prefetching sets to a temporary directory within /home, iff it is a
separate mount point
with rpe@ and deraadt@, "ffiinaallllyyy .... OK" rpe@ (r.i.p. progress bars)
djm [Fri, 17 Jan 2014 00:21:06 +0000 (00:21 +0000)]
signed/unsigned comparison warning fix; from portable
tobias [Thu, 16 Jan 2014 21:45:33 +0000 (21:45 +0000)]
Avoid size_t overflow in apprentice_map.
ok millert
tobias [Thu, 16 Jan 2014 21:41:22 +0000 (21:41 +0000)]
Avoid size_t overflow while reading /etc/resolv.conf.tail.
ok krw
brad [Thu, 16 Jan 2014 19:32:26 +0000 (19:32 +0000)]
Appease LLVM's integrated assembler. Matches the same code as it exists
for i386.
error: ambiguous instructions require an explicit suffix (could be 'cmpb', 'cmpw', 'cmpl', or 'cmpq')
ok mlarkin@
syl [Thu, 16 Jan 2014 10:36:33 +0000 (10:36 +0000)]
Remove an infinite loop in fuse_device_cleanup().
Spotted by pelikan@
Some input from stsp@
OK stsp@, pelikan@
mpi [Thu, 16 Jan 2014 10:26:21 +0000 (10:26 +0000)]
Replaces a lookup to find the link-layer address by accessing the if_sadl
member directly.
ok mikeb@
syl [Thu, 16 Jan 2014 09:31:44 +0000 (09:31 +0000)]
Add support for mknod in fuse.
OK tedu@
"it looks good to me" from guenther@
djm [Thu, 16 Jan 2014 07:32:00 +0000 (07:32 +0000)]
openssh-6.5
djm [Thu, 16 Jan 2014 07:31:09 +0000 (07:31 +0000)]
needless and incorrect cast to size_t can break resumption of
large download; patch from tobias@
jmc [Wed, 15 Jan 2014 16:07:27 +0000 (16:07 +0000)]
add -C to the man page, and adjust usage();
ok lteo
jmc [Wed, 15 Jan 2014 15:09:19 +0000 (15:09 +0000)]
for STANDARDS, not that the ability to specify a month name as a single
argument is also an extension;
ok sobrado
jmc [Wed, 15 Jan 2014 15:06:17 +0000 (15:06 +0000)]
fix SEE ALSO;
nicm [Wed, 15 Jan 2014 11:46:28 +0000 (11:46 +0000)]
Do not attempt to read .tmux.conf if we can't figure out a home
directory, from Tiago Cunha.
nicm [Wed, 15 Jan 2014 11:44:18 +0000 (11:44 +0000)]
Couple of fixes from cppcheck via Tiago Cunha.
aoyama [Wed, 15 Jan 2014 11:13:53 +0000 (11:13 +0000)]
Add WSDISPLAYIO_SETGFXMODE ioctl on LUNA's frame buffer.
This ioctl changes `pseudo' frame buffer depth, in order to use color
wscons and monochrome X server both. Also need to some
luna88k-specific initialization in
xenocara/driver/xf86-video-wsfb/src/wsfb_driver.c.
mpi [Wed, 15 Jan 2014 11:10:40 +0000 (11:10 +0000)]
Remove a hack to update the address and packet length for every device
request and instead re-open the default pipe with updated values when
attaching a new device, adapted from FreeBSD r162977.
This fixes a problem where the controller could have cached the previous
values and would fail to get the device descriptor, leaving the device
unrecognized with a message like: "device problem, disabling port n".
pirofti [Wed, 15 Jan 2014 11:01:46 +0000 (11:01 +0000)]
Add documentation for the octrtc driver.
pirofti [Wed, 15 Jan 2014 10:36:10 +0000 (10:36 +0000)]
Enable by default the DS1337 TOD clock.
pirofti [Wed, 15 Jan 2014 10:29:45 +0000 (10:29 +0000)]
Add support for the DS1337 TOD clocks found on some of the octeon models.
This is a very low resolution clock (1 second) that some models seem
to be blessed with.
Found at least on CAM-100 and DSR-500 models.
It seems the EdgeRouter doesn't have support for this.
Tested by jmatthew@ and bcallah@.
Okay miod@, bcallah@
mpi [Wed, 15 Jan 2014 09:25:38 +0000 (09:25 +0000)]
Remove assigned but never read value.
ok stsp@
deraadt [Wed, 15 Jan 2014 05:31:51 +0000 (05:31 +0000)]
remove functions that were uses by other (non-linux) compat; and
convert to ANSI protos while here
lteo [Wed, 15 Jan 2014 04:43:36 +0000 (04:43 +0000)]
Add support for a -C option. It works on a checklist like -c but only
does the checksum comparison for selected files that are specified on
the command line.
idea discussed with deraadt@ and tedu@
manpage feedback jmc@
feedback/OK deraadt@ millert@
sthen [Wed, 15 Jan 2014 02:25:34 +0000 (02:25 +0000)]
Initialise _res.nsaddr_list in res_init(), fixing some programs which depend
on bind resolver internals. Note, the list will not be updated if resolv.conf
is changed while the program runs unless the caller resets the RES_INIT flag.
Approach suggested by otto@, original diff from Kapetanakis Giannis, fix
suggested by tedu@ and dug up again by Riccardo Mottola. Looks ok to eric@.
rpe [Wed, 15 Jan 2014 00:55:21 +0000 (00:55 +0000)]
Since we now show the installer output we want that new line back.
OK halex@
espie [Wed, 15 Jan 2014 00:31:34 +0000 (00:31 +0000)]
pipes mean read/write may short out.
okay tedu@
martynas [Tue, 14 Jan 2014 22:26:30 +0000 (22:26 +0000)]
Add wcstring attribute support for Wbounded. To be used for wchar.h
which operates on element counts rather than buffer sizes. I'll start
annotating headers in a few weeks, after the hackathon. OK millert@.
tedu [Tue, 14 Jan 2014 21:34:30 +0000 (21:34 +0000)]
most common pbkdf failure is no password, so check that first.
prompted by fritjof
tedu [Tue, 14 Jan 2014 21:33:10 +0000 (21:33 +0000)]
rearrange the bottom of main so it is less of a "zoo". ok deraadt
miod [Tue, 14 Jan 2014 18:21:37 +0000 (18:21 +0000)]
build crt*S with -fPIC instead of -fpic
jmc [Tue, 14 Jan 2014 17:52:15 +0000 (17:52 +0000)]
add /etc/signify; description supplied by tedu
jmc [Tue, 14 Jan 2014 17:49:24 +0000 (17:49 +0000)]
stick that last example in a display, and avoid linewrap;
tedu [Tue, 14 Jan 2014 17:25:06 +0000 (17:25 +0000)]
this can be static
tedu [Tue, 14 Jan 2014 17:15:12 +0000 (17:15 +0000)]
stdlib.h for malloc. from Fritjof
naddy [Tue, 14 Jan 2014 13:59:45 +0000 (13:59 +0000)]
use a pipe, as intended, in the example
naddy [Tue, 14 Jan 2014 13:57:20 +0000 (13:57 +0000)]
fix public key names
espie [Tue, 14 Jan 2014 10:05:58 +0000 (10:05 +0000)]
reorder signature checks, we can do much more upfront.
clean up temp files always.
use a regexp for allowed keys, put default key up in front still.
mlarkin [Tue, 14 Jan 2014 09:57:51 +0000 (09:57 +0000)]
Typo in a printf, should be 'bytes' not 'blocks'
jmc [Tue, 14 Jan 2014 07:42:42 +0000 (07:42 +0000)]
update references to posix-2008, and point STANDARDS to some of the extension
notes listed in DESCRIPTION, for completeness sake;
ok otto sobrado
jmc [Tue, 14 Jan 2014 07:40:29 +0000 (07:40 +0000)]
sundry tweaks;
guenther [Tue, 14 Jan 2014 02:55:09 +0000 (02:55 +0000)]
No exception is made for symlinks in the spec, so use
fchmodat(AT_SYMLINK_NOFOLLOW) to set the mode on symlinks, pass
AT_SYMLINK_NOFOLLOW to utimensat(), and then let those be called
on symlinks by dropping the test that skipped them. Eliminate
set_lids() by changing set_ids() to use fchownat(AT_SYMLINK_NOFOLLOW)
and delete a redundant conditional in each of set_ftime()/fset_ftime().
suggested by espie@
ok millert@
guenther [Tue, 14 Jan 2014 02:44:57 +0000 (02:44 +0000)]
getnstr() returns KEY_RESIZE if there was a pending resize event, so loop
until it stops returning that, resetting the cursor position each time.
hint from Gregor Best (gbe (at) ring0.de)
problem noted by and ok sthen@
martynas [Tue, 14 Jan 2014 02:03:57 +0000 (02:03 +0000)]
Add a new option "-fstack-protector-strong" for GCC4. This includes
additional functions to be protected --- those that have local array
definitions, or have references to local frame addresses.
Note 1: Han explicitly licensed this under GPLv2 for us.
Note 2: Do *not* use this anywhere in "src" Makefiles, as the other
GCC doesn't have this option yet (but I'm working on it).
jeremy [Tue, 14 Jan 2014 01:02:46 +0000 (01:02 +0000)]
Add information on the new ruby21 FLAVOR, and add text about manually
checking PLISTs for gem ports with C extensions.
krw [Mon, 13 Jan 2014 23:42:18 +0000 (23:42 +0000)]
peek_token() a bit more to replace a bunch of manual checks with
the perfectly adequate parse_semi(). And some blocks didn't even
need to peek.
millert [Mon, 13 Jan 2014 23:18:57 +0000 (23:18 +0000)]
Add the "next" keyword as an alias for "+ 1" for relative times.
Also support "months" and "years" keywords when specified as relative
time units. All as per POSIX. Man page changes OK jmc@ sobrado@
millert [Mon, 13 Jan 2014 23:14:17 +0000 (23:14 +0000)]
Callers of get_state() and sole_reduction() always store the result
as a short so make the return value match the type of the value we
are returning. From Michael W. Bombardieri
bluhm [Mon, 13 Jan 2014 23:03:52 +0000 (23:03 +0000)]
Call all local valiables of type struct in6_ifaddr "ia6". This is
consistent with struct ifaddr "ifa" and struct in_ifaddr "ia".
OK mpi@
tedu [Mon, 13 Jan 2014 22:29:32 +0000 (22:29 +0000)]
update for new options and clarify
krw [Mon, 13 Jan 2014 21:36:46 +0000 (21:36 +0000)]
Don't eat another token looking for a ';' after skip_to_semi() has
been invoked. Fixes silent loss of the statement following a broken or
inapplicable 'interface' declaration in a lease, or an unrecognized
lease attribute.
krw [Mon, 13 Jan 2014 21:04:19 +0000 (21:04 +0000)]
No need to 'clear the peek buffer' when closing a file. The next
file opener will call new_parse() to initialize that amoung
other things. Nuke extraneous blank line in passing.
krw [Mon, 13 Jan 2014 20:56:24 +0000 (20:56 +0000)]
Don't eat two tokens when encountering a non-terminal '}'. Avoids
possibly ignoring entire rest of dhclient.conf or dhclient.leases.if
looking for a mistakenly consumed '}'.
espie [Mon, 13 Jan 2014 18:44:41 +0000 (18:44 +0000)]
fix another chmod on permanent tempfile to respect umask.
espie [Mon, 13 Jan 2014 18:42:34 +0000 (18:42 +0000)]
have Ustar extract owners for symlinks, which isn't really important for
pkg_add, since it skips links and restores owners from the plist anyways.