summaryrefslogtreecommitdiff
path: root/sys
AgeCommit message (Collapse)Author
2008-10-13Almost complete rewrite of iwn(4).Damien Bergamini
Add support for Intel WiFi Link 5000 Series adapters (5100/5150/5300/5350). Quite frankly, Intel made things unnecessarily difficult by gratuitously changing firmware commands (adding new fields in the middle of a struct) and some register offsets for the 5000 Series and by defining yet another way of loading a firmware. I had to write a hardware abstraction layer to manage those differences. Committed over a 5300 adapter (3T3R). Require an upgrade of the iwn-firmware package, even for 4965AGN users. Not tested on 5100, 5150 and 5350 (test reports are more than welcomed.) There should be no regression on the 4965AGN.
2008-10-13disable the pcb linking for udp for the moment since there is some weirdHenning Brauer
bug with IPv6 in some circumstances. we'll find it one day... lots of debugging dhill
2008-10-12Some more cleanup. No binary change.Marco Pfatschbacher
OK reyk@
2008-10-12Reduce diff to macppc version.Mark Kettenis
2008-10-12Adjust for redesigned powerpc interrupt architecture.Mark Kettenis
2008-10-12if converters in a group are not needed to satisfy the requirementsJacob Meuser
of a stream, direct the unneeded converters use HDA link stream #0 from NetBSD via Alexey Suslikov, thanks tested by several, thanks also
2008-10-12add azalia_codec_disconnect_stream(), a function to direct allJacob Meuser
converters in a group to use HDA link stream #0. use it when halting playback/recording. from NetBSD via Alexey Suslikov, thanks tested by several, thanks also
2008-10-12Make vcons(4) interrupt driven.Mark Kettenis
2008-10-12Add interrupt support.Mark Kettenis
2008-10-12If re_attach() fails at any point make sure to disestablish variousBrad Smith
resources setup by the bus front ends, especially the shutdown hook. Found while testing some other changes with a CardBus adapter and re_attach() was failing early before the interface pointer was assigned. Then the system was rebooted, the shutdown hook was called and *boom* in re_stop().
2008-10-11Use hardware timer to simulate interrupt moderation. Old devices will noBrad Smith
longer be livelocked when they are receiving on GigE line. Newer devices also gain well controlled interrupt rate. If hardware supports interrupt moderation (e.g. 8168B, 8168C), you could also use hardware based interrupt moderation, however, due to lack of necessary information it does not work as reliably as simulated interrupt moderation. It is _not_ recommended currently. By default, PCIe devices' simulated interrupt moderation timer is set to 75us, while PCI devices' is set to 125us. From sephe@DragonFly
2008-10-11Make sure no two pppoe devices share the same sc_unique identifier. WhileMiod Vallat
there, fix some typos, and pass M_CANFAIL to all malloc() calls which use M_WAITOK but are tested for failure. test&ok brad@
2008-10-11Fix various printf constructs either missing arguments or format specifiers.Miod Vallat
2008-10-11Since malloc_page_alloc() is a pool allocator it should check for PR_WAITOKMark Kettenis
instead of M_NOWAIT. Checking for M_NOWAIT made many malloc calls that used that flag actually wait. This probably explains many if the strange hangs people have seen recently. ok miod@
2008-10-11For none interrupt endpoints call the xfer done function beforeMarcus Glocker
calling the xfer callback function to allow a device driver to immediately reschedule the same xfer. This e.g. improves performance for isoc high- fullspeed xfers (ehci). From NetBSD. OK phessler@
2008-10-10Enable PCIe snooping on the remaining ATI/NVIDIA chipsets.Brad Smith
Fixes clicking and other noise related issues. ok ratchov@
2008-10-10Use the MI random() for all cases; no need for a MD version since itTheo de Raadt
has never been performance sensitive. Running on all platforms, discussed with millert and kettenis, ok toby
2008-10-10minor 3 was /dev/prandom, which used random(). make that use theTheo de Raadt
arandom code, just like minor 4. in fact, make them act exactly the same.
2008-10-10back out previous change. Another panic, not as frequent, andDavid Hill
definitely not at will.
2008-10-10Comment out statekey code to stop 'panic: soreceive 3', whichDavid Hill
happens with IPv6 TCP traffic, until a better fix is found. patch from henning@ proded by deraadt@
2008-10-10backout; is causing some people difficultyTheo de Raadt
2008-10-10- add and enable rum(4)Jasper Lievisse Adriaanse
committed over a rum(4) donated by Maxim Belooussov, thanks! ok jsing@
2008-10-10missing spl() for codec command setting/retrievalJacob Meuser
from NetBSD via Alexey Suslikov, thanks "looks good" ratchov@
2008-10-10Add empty cpu_unidle() macros for architectures that currently don't doArtur Grabowski
anything special to prod a cpu to leave the idle loop in signotify. powerpc, i386, amd64 and sparc64 will follow soon so that everyone has the same interface to wake an idling cpu.
2008-10-10Define MAXCPUS on all architectures.Artur Grabowski
For now, sparc64 is arbitrarily set to 256 (only architecture that didn't have a practical limit in the code on the number of cpus).
2008-10-09Set hardware timer according to bus clock speed.Brad Smith
From sephe@DragonFly
2008-10-09Skip memory areas smaller than 1 MB above the 16 MB boundary to deal withMark Kettenis
buggy ACP BIOSes. Based on a diff from weingart@. ok deraadt@, oga@
2008-10-09replace M_WAITOK with M_NOWAITCharles Longeau
ok krw@ kettenis@
2008-10-09Change sb_timeo to unsigned, so that even if some calculation (ie. n * HZ)Theo de Raadt
becomes a very large number it will not wrap the short into a negative number and screw up timeouts. It will simply become a max of 65535. Since this happens when HZ is cranked to a high number, this will still only take n seconds, or less. Safer than crashing. Prompted by PR 5511 ok guenther
2008-10-09remove a dead store.Charles Longeau
Found by LLVM/Clang Static Analyzer. ok weingart@
2008-10-09Implement CPU_INFO_UNIT for everyone, not just MP kernels.Artur Grabowski
ok miod@
2008-10-09pass NULL rather than 0 for interrupt time usage to calcru()Kevin Lo
ok art@
2008-10-09Use the reference count in struct process to update the proc countPhilip Guenthe
for users by the correct amount when changing real UID on (r)threaded processes. ok art@ tedu@
2008-10-09Put a reference count in struct process to prevent use-after-freePhilip Guenthe
if the main thread reaches the reaper ahead of some other thread in the process. ok art@ tedu@
2008-10-09Add fallback ioctl handlers for MTU and multicast to ether_ioctl().Brad Smith
2008-10-09stop waiting for the printer to become ready, see PR 5427 for moreTheo de Raadt
information. Also explains PR 4124 and 5318
2008-10-09Use ether_ioctl().Brad Smith
2008-10-09cosmetic change for ioctl funtions.. move splnet out from variable declaration.Brad Smith
2008-10-08cosmetic change for ioctl funtions.. move splnet out from variable declaration.Brad Smith
2008-10-08syncDamien Bergamini
2008-10-08new PCI ID for Intel WiFi Link 5350.Damien Bergamini
2008-10-08better commentsOleg Safiullin
2008-10-08Older IT8712F chips have 8-bit watchdog timeout counter.Oleg Safiullin
Use minutes for 16-bit values.
2008-10-08As the Option N.V. Globetrotter HSDPA Modem is not compatibleYojiro Uo
with ubsa(4), it is moved to umsm(4). ok jsg@, reported by Daniele Pilenga
2008-10-08o chip revision is 4-bit valueOleg Safiullin
o some #define's instead of digits
2008-10-08Don't extend amaps beyond what their supposed maximum. This code path isArtur Grabowski
not taken anymore, but it doesn't hurt to be correct. from NetBSD, through mickey in pr 5812 prodded by otto@
2008-10-08Get rid of the second table entry pool (pfr_kentry_pl2); we're alreadyRyan Thomas McBride
using the default interrupt handler for both, so there's no need to keep table entries created in interrupt context separate. ok henning art
2008-10-08uhci(4) at cardbus(4) seems to work (to the same extent that ohci(4) andTheo de Raadt
ehci(4) work; ie. it has the same variety of unplug bugs)
2008-10-07Kill a bunch of #ifdef freebsd/netbsd code in *_drv.c. No binary change.Owain Ainsworth
2008-10-07Kill the linux-ready negative return codes in ``shared'' code. We handleOwain Ainsworth
them wrong in several cases that i've noticed and Merging when needed is still fairly simple, anyway. This shaves another 500 bytes from an amd64 kernel due to not having to flip the sign on some things. It also stops my eyes bleeding. Tested by a few along with the last diff that went in.