Simon Schubert [Thu, 22 Feb 2007 21:02:49 +0000 (21:02 +0000)]
Enable building/installing of multiple kernels in nrelease.
Directly make use of this feature and build a NATA kernel.
Simon Schubert [Thu, 22 Feb 2007 15:50:50 +0000 (15:50 +0000)]
1:1 Userland threading stage 4.2/4:
Make signal system fully lwp-aware by splitting ksignal() in appropriate
functions. Introduce lwpsignal(), which now contains the logic of
ksignal(), but can be used to deliver a signal to a specific lwp.
Convert consumers of ksignal() to use lwpsignal() when they actually
generate a thread-specific signal.
Fully implement proc_stop() and proc_unstop().
Reviewed-by: Thomas E. Spanjaard <tgen@netphreax.net>
Simon Schubert [Thu, 22 Feb 2007 15:49:08 +0000 (15:49 +0000)]
Make setrunnable() behave more like it is intended.
Last changes made setrunnable() change lwp_stat, which is definitely
wrong. Rework the code to match the original intent.
Reviewed-by: Thomas E. Spanjaard <tgen@netphreax.net>
Simon Schubert [Thu, 22 Feb 2007 15:48:55 +0000 (15:48 +0000)]
Fix stopping of the current process, which was broken by the last commit.
If the current process is still far inside the kernel and is going to
tsleep(), CURSIG()/issignal() in tsleep() have to see the pending signal;
holding the process in SSTOP is not enough.
However, if this is happening to a process which already received a stopping
signal and thus already is in SSTOP, but which wakes up for some reason, and
then posts a SA_STOP signal to itself, we may not ignore this, in spite of
already being in SSTOP.
All this process stopping business inside CURSIG()/issignal() makes all of
this very fragile, but we will have to live with it for now.
The bug can be reproduced by doing
% cat
^Z
% ps lx|grep '[c]at'
The correct output would be something like:
4156 stop p2 DL 0:00.00 cat
The broken behavior shows with:
17417 ttybg2 p2 TL 0:00.00 cat
Reviewed-by: Thomas E. Spanjaard <tgen@netphreax.net>
Simon Schubert [Thu, 22 Feb 2007 13:15:55 +0000 (13:15 +0000)]
Fix symbol resolution for objects loaded by ELF dependencies.
We previously used the dagmembers of the specified object to resolve
a symbol. However, dagmembers is only set for objects loaded by dlsym().
The correct behavior, implemented in this patch is to instead search
the needed objects.
This should fix KDE/SSL.
Taken-from: FreeBSD
Identified-by: Kimura Fuyuki <fuyuki@hadaly.org>
Sepherosa Ziehau [Thu, 22 Feb 2007 05:17:10 +0000 (05:17 +0000)]
- Add noise floor to stats
- Include current tx rate in stats
- Remove stub radar support
- Update ath_hal(4) to 0.9.20.3:
o Add support for newer parts that do not require separate keycache
entries for tx+rx mic keys
o No more ds_vdata in tx/rx descriptors
o Split h/w tx/rx descriptor from s/w status
o As part of the descriptor split change the rate control module API
so the ath_buf is passed in to the module so it can fetch both
descriptor and status information as needed
- Replace replace h/w descriptor struct's with proper mask+shift defs for
ath(4) sample TX rate control algorithm.
- Due to the license change in ah_osdep.[ch] (now dual-BSD/GPL), move them
out of sys/contrib/dev/ath/dragonfly and put them into sys/dev/ath/ath
- Nuke the local patch for ah_osdep.c, ah_osdep.c can be modified directly.
Obtained-from: FreeBSD (sam@freebsd.org)
YONETANI Tomokazu [Thu, 22 Feb 2007 04:02:50 +0000 (04:02 +0000)]
Don't loop beyond maximum cpu ID (ncpus - 1); this probably hasn't done
any harm because the corresponding bit in smp_active_mask is zero anyway.
Simon Schubert [Wed, 21 Feb 2007 15:47:02 +0000 (15:47 +0000)]
1:1 Userland threading stage 4.1/4:
Make ksignal() aware of multiple lwps. Streamline logic.
Introduce find_lwp_for_signal(), which returns the best
lwp to deliver a signal to.
Simon Schubert [Wed, 21 Feb 2007 15:46:48 +0000 (15:46 +0000)]
1:1 Userland threading stage 2.20/4:
Unify access to pending threads with a new function, lwp_sigpend(), which
returns pending signals for the lwp, which includes both lwp-specific
signals and signals pending on the process. The new function lwp_delsig()
is used to remove a certain signal from the pending set of both process and
lwp.
Rework the places which access the pending signal list to either use those
two functions or, where not possibly, to work on both lwp and proc signal
lists.
Simon Schubert [Wed, 21 Feb 2007 15:45:37 +0000 (15:45 +0000)]
1:1 Userland threading stage 2.19/4:
Move p_sig and p_code to struct lwp and add comment for them.
Make elf coredumps and checkpointing more lwp-aware.
Pass the signal responsible for the coredump down to the service routine.
Simon Schubert [Wed, 21 Feb 2007 02:19:39 +0000 (02:19 +0000)]
Replace magic number with constant.
Spotted-by: Jason Smethers
Thomas E. Spanjaard [Tue, 20 Feb 2007 15:58:27 +0000 (15:58 +0000)]
Add NATA devices to LINT, and hook the natacontrol binary into the build,
as it does no harm even when not using NATA.
Prodded-by: Justin C. Sherrill <justin@shiningsilence.com>
Sascha Wildner <saw@online.de>
Sascha Wildner [Tue, 20 Feb 2007 15:52:43 +0000 (15:52 +0000)]
Fix typo.
Sepherosa Ziehau [Tue, 20 Feb 2007 14:24:21 +0000 (14:24 +0000)]
Don't assume mgtq will always be empty, drain mgtq explicitly.
Tested-by: Johannes Hofmann <Johannes.Hofmann@gmx.de> (iwi(4))
Sascha Wildner [Mon, 19 Feb 2007 11:10:12 +0000 (11:10 +0000)]
Fix typos.
Sascha Wildner [Mon, 19 Feb 2007 09:34:33 +0000 (09:34 +0000)]
Do some syncing with reality.
Simon Schubert [Mon, 19 Feb 2007 01:14:54 +0000 (01:14 +0000)]
Silence compiler warning.
Simon Schubert [Mon, 19 Feb 2007 01:14:40 +0000 (01:14 +0000)]
Remove outdated comment.
Simon Schubert [Mon, 19 Feb 2007 01:14:24 +0000 (01:14 +0000)]
1:1 Userland threading stage 2.18/4:
Push lwp use a bit further by making some places lwp aware.
This commit deals with ddb, procfs/ptrace and various consumers of
allproc_scan.
Sascha Wildner [Mon, 19 Feb 2007 00:51:54 +0000 (00:51 +0000)]
Remove unused malloc type M_ACL.
Simon Schubert [Sun, 18 Feb 2007 23:46:11 +0000 (23:46 +0000)]
Build gcc41 per default. This can be disabled with NO_GCC41 in make.conf.
This does not make gcc41 the default compiler. To do so, you still have to
set CCVER=gcc41 in make.conf or in your environment.
Simon Schubert [Sun, 18 Feb 2007 16:17:09 +0000 (16:17 +0000)]
1:1 Userland threading stage 2.17/4:
Consolidate consumers of p_lock to use PHOLD/PRELE. Introduce lwp_lock,
which will be used like p_lock as a "don't destruct" marker.
Simon Schubert [Sun, 18 Feb 2007 16:16:11 +0000 (16:16 +0000)]
1:1 Userland threading stage 2.16/4:
Move P_ONRUNQ to lwp_flag.
Simon Schubert [Sun, 18 Feb 2007 16:15:24 +0000 (16:15 +0000)]
1:1 Userland threading stage 2.15/4:
Drop P_ZOMBIE and make SZOMB a real state again.
Simon Schubert [Sun, 18 Feb 2007 16:13:27 +0000 (16:13 +0000)]
1:1 Userland threading stage 2.14/4:
Extend select to be usable for multiple LWPs in one proc.
Simon Schubert [Sun, 18 Feb 2007 16:12:43 +0000 (16:12 +0000)]
1:1 Userland threading stage 2.13/4:
Move P_SINTR and P_BREAKTSLEEP into lwp_flag.
Introduce proc_stop and proc_unstop to handle the transition of a complete proc
to and from stopped state. This is influenced by NetBSD.
Simon Schubert [Sun, 18 Feb 2007 14:28:18 +0000 (14:28 +0000)]
Re-order #includes to silence warning and use inline version of ``ffs''.
Sascha Wildner [Sun, 18 Feb 2007 14:17:29 +0000 (14:17 +0000)]
* Add .Nm and MLINKS for BIOS_* macros.
* Enhance markup a bit.
Sascha Wildner [Sun, 18 Feb 2007 09:09:43 +0000 (09:09 +0000)]
Fix typo.
Sascha Wildner [Sun, 18 Feb 2007 07:12:19 +0000 (07:12 +0000)]
Remove unused variable.
Sascha Wildner [Sat, 17 Feb 2007 09:53:22 +0000 (09:53 +0000)]
Uncomment NULLFS which was fixed by Matt some while ago.
Submitted-by: Thomas Nikolajsen <thomas.nikolajsen@mail.dk>
Sepherosa Ziehau [Sat, 17 Feb 2007 07:05:54 +0000 (07:05 +0000)]
- Nuke no longer needed macros.
- Use ieee80211_tim_ie, instead of local TIM ie definition.
- It turns out TIM template can contain more ies than I expected --
It is safe to put Exteneded Rate, WPA, RSN and WME ies into TIM template,
thus duplicated TIM ie in infrastructure BSS's beacon frames can be
eliminated:
o Move TIM template setup from template initalization routine to beacon
template setup routine.
o Use ies' location information provided by ieee80211_beacon_offset to put
all fixed fields and ies before TIM ie into beacon template. Ies starting
from TIM ie are put into TIM template.
Simon Schubert [Sat, 17 Feb 2007 02:24:20 +0000 (02:24 +0000)]
Update to gcc-4.1.2.
Simon Schubert [Sat, 17 Feb 2007 02:23:58 +0000 (02:23 +0000)]
Update our README.
Simon Schubert [Sat, 17 Feb 2007 02:23:28 +0000 (02:23 +0000)]
Merge from vendor branch GCC:
Import gcc-4.1.2.
Simon Schubert [Sat, 17 Feb 2007 02:23:28 +0000 (02:23 +0000)]
Import gcc-4.1.2.
Victor Balada Diaz [Fri, 16 Feb 2007 23:41:02 +0000 (23:41 +0000)]
Add a missing header that prevents detecting the kernel is compiled with INET6 support
Noticed-by: corecode@
Simon Schubert [Fri, 16 Feb 2007 23:11:40 +0000 (23:11 +0000)]
1:1 Userland threading stage 2.12/4:
Factor out lwp_stat and move P_STOPPED into p_stat.
Reviewed-by: Thomas E. Spanjaard <tgen@netphreax.net>
Simon Schubert [Fri, 16 Feb 2007 12:45:12 +0000 (12:45 +0000)]
Document and correctly name fields.
Sig now is the pending signals for a process, tsig the pending signals
for a lwp.
Remove non-existing fields from the documentation.
Prodded-by: swildner@
Sepherosa Ziehau [Fri, 16 Feb 2007 11:46:47 +0000 (11:46 +0000)]
- Clean up device header file inclusion.
- Move device sysctl creation before ieee80211_ifattach(), so if
the creation fails, there is no need to call ieee80211_ifdetach()
before calling acx_detach().
- Add radio tap support. acx100 part does not work correctly if
privacy is used, mainly due to firmware decrytion. Also acx100
part does not seem to be able to differentiate long preamble frames
from short preamble ones.
Sepherosa Ziehau [Fri, 16 Feb 2007 11:17:01 +0000 (11:17 +0000)]
Add ieee80211_plcp2rate() to convert PLCP signal/rate to net80211 rate.
This function returns 0 if unrecognized PLCP signal/rate is passed.
Sascha Wildner [Fri, 16 Feb 2007 10:11:46 +0000 (10:11 +0000)]
Fix path.
Noticed-by: Andreas Hauser
Sepherosa Ziehau [Fri, 16 Feb 2007 06:34:10 +0000 (06:34 +0000)]
- Factor out acx_set_chan(), nuke ACX_ENABLE_{TX,RX}CHAN() macros.
- Rework acx_newstate() to avoid code duplication.
- Regroup function declarations.
- Nuke no longer needed code in acx_next_scan()
- Nuke declaration of acx_beacon_intvl, it has already gone.
Simon Schubert [Fri, 16 Feb 2007 02:15:25 +0000 (02:15 +0000)]
Fix RSS display by calculating the correct value.
For some reason vmspace->vm_rssize is always zero, so we have to
"calculate" the value via vmspace_resident_count(), which actually
accesses the vmspace's pmap resident count.
Reported-by: victor@
Victor Balada Diaz [Thu, 15 Feb 2007 21:18:25 +0000 (21:18 +0000)]
Merge from vendor branch BIND:
update the readme files for bind-9.3.4 import
Victor Balada Diaz [Thu, 15 Feb 2007 21:18:25 +0000 (21:18 +0000)]
update the readme files for bind-9.3.4 import
Victor Balada Diaz [Thu, 15 Feb 2007 21:03:46 +0000 (21:03 +0000)]
Merge from vendor branch BIND:
Import bind-9.3.4
Victor Balada Diaz [Thu, 15 Feb 2007 21:03:46 +0000 (21:03 +0000)]
Import bind-9.3.4
Victor Balada Diaz [Thu, 15 Feb 2007 20:28:50 +0000 (20:28 +0000)]
Import bind-9.3.4
Sascha Wildner [Thu, 15 Feb 2007 09:20:57 +0000 (09:20 +0000)]
The KAME project has been dissolved and kame.4 isn't terribly useful.
Taken-from: FreeBSD
Sepherosa Ziehau [Thu, 15 Feb 2007 09:05:11 +0000 (09:05 +0000)]
- Add 'chan' parameter for acx_join_bss(), since for Monitor operational mode
current channel is not pointed by ieee80211com.ic_bss.ni_chan.
- Factor out acx_rx_config() from acx_write_config().
- Add Monitor operational mode support.
- Correct promisc support.
- On TX path:
o Move BPF_MTAP() before ieee80211_encap(), so ether frames will be tapped.
o Tap 802.11 MAC frames to ieee80211com.ic_rawbpf before they are encrypted.
- On RX path:
o Tap ACK/CTS to ieee80211com.ic_rawbpf.
o For non-Monitor operational mode, update ieee80211_stats.is_rx_tooshort,
if frames are too short.
- Nuke unused functions.
Sepherosa Ziehau [Wed, 14 Feb 2007 13:00:34 +0000 (13:00 +0000)]
PCIe re(4) can't handle TCP csum offloading well if short packets
are padded by the driver, which is intended to fix PCI re(4) csum
offloading bug. It turns out both PCI and PCIe re(4) _can_ handle
short packets TCP csum offloading without driver's interferece, so
padding for short TCP packets is avoided.
Obtained-from: FreeBSD (wpaul@freebsd.org)
Tested-by: Joe Talbott <josepht@cstone.net> RTL8101E(PCIe)
me RTL8169S(PCI) RTL8169SB(PCI)
Sepherosa Ziehau [Tue, 13 Feb 2007 14:15:57 +0000 (14:15 +0000)]
Reclaim challenge text, if we operate as an AP and other side switches auth mode
Obtained-from: Atheros via FreeBSD (sam@freebsd.org)
Sepherosa Ziehau [Mon, 12 Feb 2007 15:02:33 +0000 (15:02 +0000)]
- Replace 0 with NULL when pointers are tested.
- Break long line.
- Style(9) for 'if' statement.
- Correct bpf_freed() function comment.
Sascha Wildner [Mon, 12 Feb 2007 14:54:39 +0000 (14:54 +0000)]
Sync zoneinfo database with tzdata2007b from elsie.
leapseconds: 8.1 -> 8.2
Change the commentary to reflect the IERS announcement that there is to be
no positive leap second at the end of June 2007.
Sepherosa Ziehau [Mon, 12 Feb 2007 14:05:13 +0000 (14:05 +0000)]
- Prefer m_getl()
- Nuke blank line
Sepherosa Ziehau [Sun, 11 Feb 2007 12:10:04 +0000 (12:10 +0000)]
- Rename reset_d() to bpf_resetd() just like other function which operates
on bpf_d
- Use m_lengthm()
- Nuke tab on blank line
Sepherosa Ziehau [Sun, 11 Feb 2007 11:30:05 +0000 (11:30 +0000)]
In bpf_ptap(), duplicate incoming mbuf's m_pkthdr.rcvif, so !bpf_d.bd_seesent
will be handled correctly in bpf_mtap().
Sepherosa Ziehau [Sun, 11 Feb 2007 10:23:29 +0000 (10:23 +0000)]
kmalloc(M_WAITOK) will not return NULL
Sascha Wildner [Sun, 11 Feb 2007 01:51:28 +0000 (01:51 +0000)]
Remove ancient SimOS support.
Sascha Wildner [Sun, 11 Feb 2007 01:07:19 +0000 (01:07 +0000)]
Fix function name.
getosreldate(3) should probably be nuked entirely.
Sascha Wildner [Sun, 11 Feb 2007 00:47:01 +0000 (00:47 +0000)]
Fix function names.
Sascha Wildner [Sun, 11 Feb 2007 00:13:18 +0000 (00:13 +0000)]
Remove MCHTYPE() documentation (removed in rev. 1.14 of sys/sys/mbuf.h).
Sascha Wildner [Sat, 10 Feb 2007 23:54:18 +0000 (23:54 +0000)]
Remove documentation of extinct macro MULTI_DRIVER_MODULE() (nuked in
rev. 1.11 of sys/sys/bus.h).
Sascha Wildner [Sat, 10 Feb 2007 21:04:12 +0000 (21:04 +0000)]
Fix typo: cgetseq -> cgetset
Sascha Wildner [Sat, 10 Feb 2007 20:46:36 +0000 (20:46 +0000)]
Remove documentation of the recently removed fuswintr() and suswintr()
functions.
Sepherosa Ziehau [Sat, 10 Feb 2007 11:32:39 +0000 (11:32 +0000)]
- Move vke_softc.sc_kqueue initialization from vke_start() into vke_init()
- Add missing BPF_MTAP() in vke_start()
Sepherosa Ziehau [Sat, 10 Feb 2007 07:03:45 +0000 (07:03 +0000)]
Describe new added rum(4) ids.
Obtained-from: OpenBSD
Sepherosa Ziehau [Sat, 10 Feb 2007 05:45:12 +0000 (05:45 +0000)]
Add more ids
Obtained-from: OpenBSD
Sepherosa Ziehau [Sat, 10 Feb 2007 05:44:19 +0000 (05:44 +0000)]
regen
Sepherosa Ziehau [Sat, 10 Feb 2007 05:43:08 +0000 (05:43 +0000)]
More rum(4) ids
Sepherosa Ziehau [Fri, 9 Feb 2007 11:31:41 +0000 (11:31 +0000)]
Avoid possible mbuf re-tap on error path.
Thomas E. Spanjaard [Thu, 8 Feb 2007 21:48:24 +0000 (21:48 +0000)]
Misc cleanups to NATA ata-raid.c to make it compile. Fix malloc/free ->
kmalloc/kfree, M_NOWAIT -> M_WAITOK, and various bugfixes and warnings
fixes. Note, that this hasn't been tested in actual operation, that's
something for next week when I have the time + hardware.
Sepherosa Ziehau [Thu, 8 Feb 2007 15:39:39 +0000 (15:39 +0000)]
Convert RX buffer signal level to relative RX signal strength (relative to
noise floor).
The conversion for acx100 part is based on Linux acx100 driver.
The conversion for acx111 part is worked out by comparing RSSI of ath(4)/ral(4)
under the same conditions with the acx(4) RX buffer signal level.
Simon Schubert [Thu, 8 Feb 2007 10:18:21 +0000 (10:18 +0000)]
Use proc flags instead of lwp flags.
Noticed-by: YONETANI Tomokazu <qhwt+dfly@les.ath.cx>
Simon Schubert [Wed, 7 Feb 2007 16:35:25 +0000 (16:35 +0000)]
Use a new pkgsrc bootstrap to work with recent tar changes.
Simon Schubert [Wed, 7 Feb 2007 16:34:32 +0000 (16:34 +0000)]
Do not create /var/db/pkg, pkgsrc bootstrap doesn't like it.
Sepherosa Ziehau [Wed, 7 Feb 2007 14:52:42 +0000 (14:52 +0000)]
Convert RX desc's rssi field to relative RX signal strength (relative to
noise floor).
The conversion for Maxim, Philips and GCT RF are adapted from Realtek's
818x Linux driver.
Sepherosa Ziehau [Wed, 7 Feb 2007 12:34:26 +0000 (12:34 +0000)]
For rum(4) and rt2661 part of ral(4):
Return -1 in {rum,rt2661}_get_rssi(), if rssi can't be converted to relative
RX signal strength. Caller of these two functions will detect the conversion
error and use old rssi to feed ieee80211_input(), so when rssi smoothing is
in place, rssi conversion error will have no impacts on average rssi.
Thomas E. Spanjaard [Tue, 6 Feb 2007 15:17:44 +0000 (15:17 +0000)]
Misc cleanups for NATA ataraid, and write ata_raid_dump() to look like
something that will actually work. I did not take the same approach as with
ad_dump() here, because that takes just too much code duplication between
ata_raid_strategy() and ata_raid_dump().
Sepherosa Ziehau [Tue, 6 Feb 2007 14:40:32 +0000 (14:40 +0000)]
Absolute RX siganl strength correction value from EEPROM should be within
[-10, 10], if it is out of range, reset it to zero.
This range is referenced from Linux RT61 driver.
Sepherosa Ziehau [Tue, 6 Feb 2007 14:33:39 +0000 (14:33 +0000)]
Correctly calculate absolute RX signal strength and use relative RX signal
strength (relative to noise floor) as RSSI. This makes the signal of node
reported by ifconfig(8) consistent with reality, i.e. weaker signal, lower
RSSI.
Calculation of absolute RX signal strength is adapted from Linux RT73 driver.
Sepherosa Ziehau [Tue, 6 Feb 2007 13:50:25 +0000 (13:50 +0000)]
Use relative RX signal strength (relative to noise floor) as RSSI.
RSSI correction value is obtained from Linux RT25USB driver.
Sepherosa Ziehau [Tue, 6 Feb 2007 12:46:09 +0000 (12:46 +0000)]
For rt2661 part:
Use relative RX signal strength (relative to noise floor) as RSSI.
Sepherosa Ziehau [Tue, 6 Feb 2007 12:38:30 +0000 (12:38 +0000)]
Use relative RX signal strength (relative to noise floor) as RSSI.
YONETANI Tomokazu [Tue, 6 Feb 2007 05:56:03 +0000 (05:56 +0000)]
leftover from 1:1 Userland threading stage 2.11/4
Simon Schubert [Mon, 5 Feb 2007 22:19:28 +0000 (22:19 +0000)]
Add flag to optionally disable stripping of old kernel and modules.
Submitted-by: Thomas Nikolajsen <thomas.nikolajsen@mail.dk>
Thomas E. Spanjaard [Mon, 5 Feb 2007 18:10:14 +0000 (18:10 +0000)]
Make natadisk's ad_dump() actually work. Ditch the fake bio/buf scheme to
pass information to ad_strategy() in favour of calling ata_queue_request()
directly. This way, we avoid hairy bio stuff and request callbacks.
Sepherosa Ziehau [Mon, 5 Feb 2007 15:19:04 +0000 (15:19 +0000)]
Convert RSSI to receive signal strength (dbm). The real convertion is not yet
hooked, but it will be used when roaming/background scan is imported.
EEPROM offset of RSSI mapping value, default RSSI mapping value and noise floor
are obtained from Linux RT2500 driver.
Sepherosa Ziehau [Mon, 5 Feb 2007 10:53:18 +0000 (10:53 +0000)]
Nuke unused header files
Simon Schubert [Mon, 5 Feb 2007 09:38:19 +0000 (09:38 +0000)]
Don't hand-roll an own version on selrecord. This unbreaks LINT build.
Reported-by: swildner@
Peter Avalos [Sun, 4 Feb 2007 21:17:34 +0000 (21:17 +0000)]
Remove unnecessary CFLAGS and SRCS lines.
Obtained-from: FreeBSD
Peter Avalos [Sun, 4 Feb 2007 21:08:28 +0000 (21:08 +0000)]
Sync with FreeBSD.
Peter Avalos [Sun, 4 Feb 2007 21:06:34 +0000 (21:06 +0000)]
Remove bogus "-" operand from usage message. There is no restriction on
where it may be placed, it is treated the same as the file arguments.
Obtained-from: FreeBSD
Simon Schubert [Sun, 4 Feb 2007 20:28:21 +0000 (20:28 +0000)]
Use the saved ucontext pointer and not the return value of get_mcontext.
Reported-by: pavalos@
Simon Schubert [Sun, 4 Feb 2007 19:45:24 +0000 (19:45 +0000)]
Don't rely on sys/resource.h including all necessary headers.
This fixes compilation on 1.4-RELEASE.
Peter Avalos [Sun, 4 Feb 2007 19:27:58 +0000 (19:27 +0000)]
Avoid truncating the pathname to UNIX Domain Sockets with snprintf(),
giving a more sensible warning when the (relatively meagre) sun_path
limit is exceeded.
Obtained-from: FreeBSD
Peter Avalos [Sun, 4 Feb 2007 18:43:39 +0000 (18:43 +0000)]
Print the "$" symbol on blank lines when the -s and -e options are used
together.
Obtained-from: FreeBSD
Matthew Dillon [Sun, 4 Feb 2007 17:24:56 +0000 (17:24 +0000)]
Add note on using 'handle SIGSEGV noprint' when gdb'ing a virtual kernel.
Suggested-by: "Simon 'corecode' Schubert" <corecode@fs.ei.tum.de>
Sepherosa Ziehau [Sun, 4 Feb 2007 13:43:53 +0000 (13:43 +0000)]
regen