summaryrefslogtreecommitdiff
path: root/sys
AgeCommit message (Collapse)Author
2002-08-08call setcontext{4,4m} directly when we know the CPU type alreadyJason Wright
2002-08-08socket buf speedup from thorpej@netbsd, okay art@ ericj@:Niels Provos
Make insertion of data into socket buffers O(C): * Keep pointers to the first and last mbufs of the last record in the socket buffer. * Use the sb_lastrecord pointer in the sbappend*() family of functions to avoid traversing the packet chain to find the last record. * Add a new sbappend_stream() function for stream protocols which guarantee that there will never be more than one record in the socket buffer. This function uses the sb_mbtail pointer to perform the data insertion. Make TCP use sbappend_stream(). On a profiling run, this makes sbappend of a TCP transmission using a 1M socket buffer go from 50% of the time to .02% of the time. Thanks to Bill Sommerfeld and YAMAMOTO Takashi for their debugging assistance!
2002-08-08th_flags doesn't have to be equal to TH_SYN to generate modulator, it'sDaniel Hartmeier
sufficient if TH_SYN is set and TH_ACK is unset, ignore TH_ECN etc. ok frantzen@
2002-08-08Use & to test if bits are set, not &&; art@ ok.Aaron Campbell
2002-08-08Completely rework transmit interrupt handling:Jason Wright
- try to handle tx ring reclaim in bestart() - if we pass over the high water mark, enable TX interrupts - disable them again when the low water mark is passed Thanks to mickey@ for enduring 4 versions of this =)
2002-08-07return EBUSY in processreadyNiels Provos
2002-08-07fix the stupidest alloc/free piece of code ever; nate@ okMichael Shalayeff
2002-08-07consistently check byte order of ether_type; pointed out by dhartmeiJason Wright
2002-08-07KNF (parenthesized returns)Jason Wright
2002-08-07Hrmph, part of the logic was correct: turn off if_timer when all descriptors ↵Jason Wright
are found to be empty.
2002-08-07fix if_timer handling:Jason Wright
- only set it to non-zero when at least on packet is enqueued - zero it if we tx'd at least one packet
2002-08-06be sure to unmap/deallocate buffers when hme_stop() is calledJason Wright
2002-08-06catch 'break' and handle appropriately, I thought I had committed this long ago.Jason Wright
2002-08-06Set RI_CLEAR if the card isn't the console framebuffer; based on discussion ↵Jason Wright
with miod.
2002-08-06No, powerpc is not Elf64, it is Elf32. stupid comment error.Dale Rahn
2002-08-05Force acceleration to be turned off if we encounter an old hardwareMiod Vallat
revision. Jason and I think that only someone out of his mind would plug an old dusty dual slot cgsix in an ultrra, but it's better to be safe anyways. ok jason@
2002-08-05Clean up hook cookies from shutdown hook.Thierry Deval
2002-08-05move HoldScreen to R3 where it belongs (not L3)Jason Wright
2002-08-05the supported agp chipset list is md-configured, not miMichael Shalayeff
2002-08-03in fact, no need for switch_exit() completely, cpu_exit() handles the reins ↵Michael Shalayeff
to the cpu_switch and basta!
2002-08-03reorder stack_alloc() and add a commentMichael Shalayeff
2002-08-03store md_regs ptr before interrupts are enabled on syscall, do not store the ↵Michael Shalayeff
frame pointer on the stack on traps, just offset back from sp upon return
2002-08-03some more opt and diagnostic checks in set/remrunqueueMichael Shalayeff
2002-08-03opt {set,rem}runqueue a littlebitMichael Shalayeff
2002-08-03no need for a stack on exit() since u gets freed in a different context, ↵Michael Shalayeff
also s/P_MD/P_MD_REGS
2002-08-03u could get a little higherMichael Shalayeff
2002-08-03rework boot() w/o a gotoMichael Shalayeff
2002-08-03we do not use _c_label() evenMichael Shalayeff
2002-08-03Don't depend on curproc, pass the proc * down to the functions that need it ↵Jason Wright
(fmovr/fmovcc).
2002-08-03spell 'debugging' correctly with double 'g' (in comments)Peter Valchev
2002-08-03- don't use unexistant as(1) .weakext pseudo; found by todd@Federico G. Schwindt
- remove unused code. mickey@ ok. libc cross-compiling (i386->hppa) reported to work now.
2002-08-02More possible int overflows found by Silvio Cesare.Todd C. Miller
ibcs2_stat.c one OK by provos@
2002-08-02registers should be volatile, also add yet more register defnsJason Wright
2002-08-02Do correct bounds checking in get/set/put cmap routines. A few ofTodd C. Miller
these check were already OK but have been modified for consistency. Problem found by Silvio Cesare.
2002-08-02In hpux_sys_getcontext(), check for len <= 0 and return EINVAL.Todd C. Miller
Noted by Silvio Cesare
2002-08-02add string for cputype (bad art =)Jason Wright
2002-08-02whitespaceJason Wright
2002-08-02prevent integer overflow by verifying that ua.num is not too large.Nathan Binkert
ok niels, millert, art
2002-08-01Fix handling of 0-length RPC fragmentsConstantine Sapuntzakis
2002-08-01whoops, last write tracker should be updated on every write to avoid the ↵Jason Wright
burst write problem on 7811.
2002-08-01add support for mapping interrupts.Jason Wright
2002-08-01Limit max file size based on PAGE_SIZE. Even though ffs can handleTodd C. Miller
files up to 16TB, we do limit the max file to 2^31 pages to prevent overflow of a 32-bit unsigned int. The buffer cache has its own checks but a little added paranoia never hurts. Adapted from a patch in FreeBSD.
2002-07-31fake out ELF stuffTheo de Raadt
2002-07-31lucifier is technical now; ok mickey@Todd T. Fries
2002-07-31Grammar.Thomas Nordin
2002-07-31de-registerJason Wright
2002-07-31Temporarily comment ELF support in the bootblocks, until this port isMiod Vallat
really ELF; necessary for this to compile after the recent <sys/exec_elf.h> changes. ok deraadt@
2002-07-31aaron: sync THEN access =)Jason Wright
2002-07-31Remove mickey's debugging printf goop.Aaron Campbell
2002-07-31init pm_ptphint, hint from art@Michael Shalayeff
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-30 23:28:07 +0000