eephy(4) diff: add Marvell PHYG65G Gigabit PHY

2010-11-21 Thread Kevin Lo
This diff from FreeBSD adds Marvell PHYG65G Gigabit PHY which is found on 88E8059 Yukon Optima. Tested by Frans Haarman. # dmesg | grep msk mskc0 at pci3 dev 0 function 0 "Marvell Yukon 88E8059" rev 0x11, Yukon-2 Optima (0x1): apic 1 int 19 (irq 11) msk0 at mskc0 port A: address 1c:c1:de:9c:04:e2

Re: usb mem wait

2010-11-21 Thread Jacob Meuser
On Sun, Nov 21, 2010 at 06:47:01PM -0500, Ted Unangst wrote: > instead of faking an assertwaitok check, let's use the real thing. this > is almost the opposite of progress on the whole bluetooth issue, but it > shortens the stack trace considerably. the curproc check isn't terribly > accurate,

Re: zap stats and ioctl from rnd

2010-11-21 Thread Theo de Raadt
> On 11/22/10 01:16, Ted Unangst wrote: > > is any of this useful? has anybody ever manually stirred the random > > device or tried interpreting the nonsense spit out by sysctl > > We stir it in the installer, FWIW. Using the xxxwrite() entry point.

Re: wihap_data_input -> wi_start -> panic: m_copydata: len -7 < 0

2010-11-21 Thread Kenneth R Westerback
On Sun, Nov 21, 2010 at 09:44:12AM -0800, Greg Steuck wrote: > On Sat, Nov 20, 2010 at 4:36 AM, Kenneth R Westerback > wrote: > > Is it as simple as a missing m_pullup(), which is apparently what > > ieee8011_encap() does in the situation? > > Looks very logical. I just applied this and booted th

Re: tar -N for numeric

2010-11-21 Thread Ted Unangst
On Sun, Nov 21, 2010 at 7:29 PM, Alexander Hall wrote: > On 11/21/10 18:15, Ted Unangst wrote: >> On Sat, 20 Nov 2010, Ted Unangst wrote: >> >>> this diff adds a -N flag to turn off this behavior, -N for numeric (or no >>> names, if you prefer). i found a similar --long-option-only in gnu tar. >>

cd(4) can be T_FIXED too

2010-11-21 Thread Kenneth R Westerback
It had to happen. A USB CD device has been found that 'forgets' to mention that it is a removable device. PR #6513. Therefore tweak the cd recognition pattern to accept T_FIXED cdrom and worm devices. Can anybody think of any reason to not do this? Submitter should be testing this as I write to m

Re: zap stats and ioctl from rnd

2010-11-21 Thread Alexander Hall
On 11/22/10 01:35, Alexander Hall wrote: > On 11/22/10 01:16, Ted Unangst wrote: >> is any of this useful? has anybody ever manually stirred the random >> device or tried interpreting the nonsense spit out by sysctl > > We stir it in the installer, FWIW. Hmmm... For my definition of "stirr", w

Re: zap stats and ioctl from rnd

2010-11-21 Thread Alexander Hall
On 11/22/10 01:16, Ted Unangst wrote: > is any of this useful? has anybody ever manually stirred the random > device or tried interpreting the nonsense spit out by sysctl We stir it in the installer, FWIW.

Re: tar -N for numeric

2010-11-21 Thread Alexander Hall
On 11/21/10 18:15, Ted Unangst wrote: > On Sat, 20 Nov 2010, Ted Unangst wrote: > >> this diff adds a -N flag to turn off this behavior, -N for numeric (or no >> names, if you prefer). i found a similar --long-option-only in gnu tar. > > slightly improved. should now work for creating archives

zap stats and ioctl from rnd

2010-11-21 Thread Ted Unangst
is any of this useful? has anybody ever manually stirred the random device or tried interpreting the nonsense spit out by sysctl kern.random? [this would also delete the rndioctl.h header entirely.] Index: dev/rnd.c === RCS file:

usb mem wait

2010-11-21 Thread Ted Unangst
instead of faking an assertwaitok check, let's use the real thing. this is almost the opposite of progress on the whole bluetooth issue, but it shortens the stack trace considerably. the curproc check isn't terribly accurate, either, so it misses bugs. i don't see any reason to avoid sleeping

small cleanup xf86-input-mouse

2010-11-21 Thread Alexandr Shadchin
Now SunMouse works through wsmouse(sunms add miod@ 20 May 2009) so we can remove our patch to SunMouse for xf86-input-mouse. Right? OK? -- Alexandr Shadchin Index: driver/xf86-input-mouse/src/mouse.c === RCS file: /cvs/xenocara/dri

Re: tar -N for numeric

2010-11-21 Thread Ted Unangst
On Sat, 20 Nov 2010, Ted Unangst wrote: > this diff adds a -N flag to turn off this behavior, -N for numeric (or no > names, if you prefer). i found a similar --long-option-only in gnu tar. slightly improved. should now work for creating archives as well, and some man page improvemnts. Index

Re: wihap_data_input -> wi_start -> panic: m_copydata: len -7 < 0

2010-11-21 Thread Greg Steuck
On Sat, Nov 20, 2010 at 4:36 AM, Kenneth R Westerback wrote: > Is it as simple as a missing m_pullup(), which is apparently what > ieee8011_encap() does in the situation? Looks very logical. I just applied this and booted the new kernel. Of course it always took days before it fired, so I won't k

Re: sys/types.h in sys/socket.h

2010-11-21 Thread Ted Unangst
Is this enough to compile socket.h by itself? I think sockcred needs to be hidden inside bsd visible too, because you aren't providing uid_t that i can see. On Sun, Nov 21, 2010 at 3:45 AM, Philip Guenther wrote: > > ...and here's a diff for discussion that removes the sa_family_t and > socklen_

Re: apmd setperf on MP machines

2010-11-21 Thread Rune Lynge
On Thu, Oct 21, 2010 at 7:36 PM, Laurence Tratt wrote: > 18 months ago I posted a patch to make apmd's -C and -A modes work > half-sensibly on multi-processor machines: > > http://marc.info/?l=openbsd-tech&m=123315164930014&w=2 > > The patch went into the tree but was backed out because on some v

X Международная Конференция “Исследование, разрабо =?UTF-8?B?0YLQutCwINC4INC/0YDQuNC80LXQvdC10L3QuNC1INCS0YvRgdC+0LrQuNGFINCi0LXRhdC90L7Qu9C+0LPQuNC4

2010-11-21 Thread Organizing Committee
X PP5P6P4QP=P0QP>P4P=P0Q P=P0QQP=P>-P?QP0P:QP8QP5QP:P0Q P:P>P=QP5QP5P=QP8Q bPQQP;P5P4P>P2P0P=P8P5, QP0P7QP0P1P>QP:P0 P8 P?QP8PP:P8Q P"P5QP=P>P;P>P3P8P8 P2 P?QP>PQQP8b 9-11 P4P5P:P0P1QQ 2010, P3. P!P0P=P:Q-PP5QP5QP1QQP3, P P>QQP8Q www.htfr.org, i...@htfr.org P#P2P0P6P0P

Re: sys/types.h in sys/socket.h

2010-11-21 Thread Philip Guenther
...and here's a diff for discussion that removes the sa_family_t and socklen_t typedefs from sys/types.h, transferring them to sys/socket.h and duping them in netinet/in.h, netinet6/in6.h, and sys/un.h as necessary. There are only two user-land fixes necessary for a build with this, in sasyncd