summaryrefslogtreecommitdiff
path: root/sys
AgeCommit message (Collapse)Author
2003-06-06Fix a copied constant, mention why a number doesnt match linux, removeDale Rahn
duplicated line.
2003-06-06Add support for Intel ICH5/ICH5R IDEAlexander Yurchenko
2003-06-06There is no point in flushing the data cache all the time. It just createsHenric Jungheim
a lot of useless ecache traffic. testing by many ok jason@
2003-06-06Identify Pentium M CPUAndreas Gunnarsson
Recognize more feature flags deraadt@ ok
2003-06-06avoid M_WAIT in network stack; netbsd revision 1.13Jun-ichiro itojun Hagino
2003-06-06don't use M_WAIT in network stack. use M_PREPEND to keep m_pkthdr.Jun-ichiro itojun Hagino
(netbsd revision 1.7)
2003-06-06don't try to forward multicast packet to mif that went away; kameJun-ichiro itojun Hagino
2003-06-06initialize blocksize from the trigger_input routine; fixes a div by zeroFederico G. Schwindt
error. problem and patch by Gabriel Kihlman in PR/2856. millert@ ok.
2003-06-05untils this is sync'ed, correctly set mixer_class for CMPCI_OUTPUT_CLASS.Federico G. Schwindt
2003-06-05Cleanup and enable kauaiata DMA. Tables should be mostly saneDale Rahn
Please _TEST_.
2003-06-05Faster ramdac operation, by reducing the delay between register writes,Miod Vallat
according to some Power9x00 documentation.
2003-06-05Cleanup:Alexander Yurchenko
- don't include both sys/types.h and sys/param.h - spaces vs tab - kill trailing spaces ok drahn@
2003-06-053/4 knockout for peter galbavyTheo de Raadt
2003-06-05ATA100 driver for apple's new machines, aka kauai.Dale Rahn
This is working with udma on a 12" PBG4, but dma is disabled for now, too many hardcoded values. 'wdc...flags 0x1' will enable it.
2003-06-04The IIi/e chips have a DMA sync register that we have been ignoring. ThisHenric Jungheim
could lead to nasty races between PIO and DMA. tested by krw@ ok jason@
2003-06-04Delete UCB term 3. When there are other copyright owners, we affirmTheo de Raadt
that we can see no legal situation under which they could require that term to remain (ie. This is equivelant to us taking the old UCB file, removing term 3 as specified by UCB, then re-applying each diff afterwards from the various authors)
2003-06-04mop up some more 3/4 license issuesTheo de Raadt
2003-06-043/4 cleanup for aaronTheo de Raadt
2003-06-04clean up my copyright so it matches the templateNathan Binkert
2003-06-04Remove the advertising clause in the UCB license which BerkeleyTodd C. Miller
rescinded 22 July 1999. This got missed in the omnibus commit.
2003-06-04fix some 3/4 for tobyTheo de Raadt
2003-06-04more term 3 & 4 from various developers, with permission of courseTheo de Raadt
2003-06-04nuke term 3 based on terms fileJason Wright
2003-06-04nuke clause 3 & 4 (ok with patrik)Jason Wright
2003-06-04Use bus_untmp() instead of rolling an inline version of it;Miod Vallat
tested and ok jason@
2003-06-04strlcpy; tested by nick@Miod Vallat
2003-06-04another 3/4Theo de Raadt
2003-06-04more 3/4 cleanups for lazy developersTheo de Raadt
2003-06-04more term 3 and 4 changesTheo de Raadt
2003-06-03terms 3 & 4 cleanup based on "terms" fileTheo de Raadt
2003-06-03fix various 3/4 licenses according to "terms" fileTheo de Raadt
2003-06-03remove some term 3 & 4 for niklasTheo de Raadt
2003-06-03three four killsMichael Shalayeff
2003-06-03three four killMichael Shalayeff
2003-06-03Make minaddr sane before calling uvm_km_suballoc()Dale Rahn
2003-06-03Play it safe - when invalidating the tsb, only touch the tag andArtur Grabowski
always set it to an invalid tag. To actually have an invalid tag make sure that we never allocate the last context in ctx_alloc. Move all that magic into a simple tsb_invalidate inline function. Fixes weird race conditions in memory handling where the page in the tlb didn't match the page tables. After discussion with henric@. Tested for a while and noone objected.
2003-06-03What is it up with you sparc64 people. Do you like sitting on criticalTheo de Raadt
fixes?
2003-06-03license cleanup regarding term 4Theo de Raadt
2003-06-03Fastpath for userland crypto requests. This change makes userlandBob Beck
crypto requests attempt to call the crypto driver directly to process crypto layer requests, as opposed to queueing them in the kernel crypto thread. If we can't use the crypto devices (i.e. they're busy) we fall back to queueing the request up in the crypto thread as before. This does allow for faster performance in some cases (smaller requests, how small seems to be dependent on the card/cpu combination) where context switching is a major issue in performance. ok deraadt@ jason@
2003-06-03move some prototypes to pfvar.h. needed soon.Henning Brauer
pf_tagname2tag, pf_tag2tagname, pf_tag_unref, pf_tag_packet
2003-06-03provide pf_tag2tagname which is the reverse of pf_tagname2tag. needed soon.Henning Brauer
2003-06-03Two term license with approval from nordin@Artur Grabowski
2003-06-03remove assumption on mbuf pointed to by m0; sync w/kameJun-ichiro itojun Hagino
2003-06-03remove unneeded break after return. sync w/kameJun-ichiro itojun Hagino
2003-06-03Use an ISC-tyle license for all my code; it is simpler and more permissive.Todd C. Miller
2003-06-03nuke clause 3 & 4 (with henric's permission)Jason Wright
2003-06-03kill clause 3 and 4 from several of my copyrights, cleanup.Dale Rahn
2003-06-03license cleaning.Artur Grabowski
2003-06-03Convert to 2 term license; OK with art@Todd C. Miller
2003-06-03Nuke terms 3 & 4 on some of my software.Tobias Weingartner