summaryrefslogtreecommitdiff
path: root/sys/arch/i386/include
AgeCommit message (Collapse)Author
2004-07-14setperf driver for powernow in amd k7 cpus. derived from a diff toTed Unangst
tech-i386@netbsd by martin vegiard, and adapted to fit in here. as yet untested. ok deraadt@
2004-07-02Maintain %f and %gs over traps. Mostly from NetBSD. Preparation for SMPNiklas Hallqvist
speedups.
2004-06-28Use new event counter API for interrupt counting on i386. deraadt@ tholo@Aaron Campbell
drahn@ millert@ ok
2004-06-25Instead of accessing ci_dev (that's an MI field), provide a marco thatArtur Grabowski
translates a cpu_info structure into a human-readable cpu number. drahn@ deraadt@ ok
2004-06-25Add extended CPUID flag value for AMD64 LONG identification. Taken fromTom Cosgrove
arch/amd64/include/specialreg.h. "fine by me" weingart@
2004-06-23Renumber I/O APICs who clash with other APICs. Also clear the deliveryNiklas Hallqvist
mode before setting it to fixed, since the MP BIOS may have left something in that field. This fixed a Quad Xeon we have been loaned over the c2k4.
2004-06-23Enter cdboot, a CD-specific second-stage bootrap.Tom Cosgrove
Testing krw@ and todd@, thanks. assistance, testing and ok weingart@
2004-06-21Put back the moving of schedstate_percpu into sched.h. This time exposeArtur Grabowski
it to userland so that i386 builds (other architectures didn't show the problem). deraadt@ ok
2004-06-21Make 'machine ddbcpu' work, and use CPU device number instead of APIC idAndreas Gunnarsson
KNF lesson and ok from niklas@
2004-06-20nope, tree breakage in libpthread. too tough to run a make build?Theo de Raadt
2004-06-20Move schedstate_percpu into sched.h so that we don't have to includeArtur Grabowski
proc.h in cpu.h on __HAVE_CPU_INFO architectures. cpu.h is usually included in param.h. This also removes the horrible kludge with ifdef SYS_PROC_H in sched.h by simply converting the inline functions into macros. With a few suggestions from nordin@ deraadt@ ok
2004-06-16We need __splbarrier to compile kernel, tho it should beAlexander Yurchenko
replaced with a real gcc barrier. ok deraadt@
2004-06-15first parts of how C3 Esther will be handled; ok tomTheo de Raadt
2004-06-13debranch SMP, have funNiklas Hallqvist
2004-06-06Don't touch any hardware registers while fetching hw.cpuspeed andAlexander Yurchenko
just return current pentium_mhz value. Update this value in all hw.setperf hoos either via its own private methods or using global update_cpuspeed hook, if registered. Also implement update_cpuspeed hook for Pentium 3. Tested by millert@, Gabriel Kihlman <gk@stacken.kth.se> and me on various i386 machines. ok tedu@
2004-05-23workaround gcc brokenness by outlining spl functions.Ted Unangst
finally solves vfs corruption. hint and ok art@, ok deraadt@
2004-05-20Properly flush instruction cache for ptrace(PT_WRTIE_{DI}, ...) on powerpcMark Kettenis
and m68k. ok drahn@, millert@
2004-05-04Move _bus_space_unmap() to machdep.c so it can be used notAlexander Yurchenko
only by the rbus code. Also knf it and sync with bus_space_unmap(). ok millert@
2004-04-19remove the GATEWAY junk thingy; at the same time, select a new fatTheo de Raadt
cluster value that noone will really need to crank.
2004-04-02K6-2/3 powernow driver. not without quirks, but mostly working.Ted Unangst
testing by david@ ok deraadt@
2004-02-19- split intel686_cpu_setup() into two parts: common for familyAlexander Yurchenko
0x6 and 0xf and only for family 0x6 - use intel686_p4_cpu_setup() for family 0xf cpus - msr-based hw.cpuspeed for pentium 4 - use pentium_mhz for hw.cpuspeed if there's nothing better (diff from tedu@) Ok tedu@
2004-02-14Simplify hw.{cpuspeed,setperf} api moving all the sysctl stuffAlexander Yurchenko
from the underlying callbacks. Testing hppa mickey@, ppc drahn@ Ok markus@ tedu@
2004-02-06some cpus start to have a NX oriented PGEX_ITheo de Raadt
2004-02-052 new cpuid ecx features in prescottTheo de Raadt
2004-02-03move VIA xcrypt-* options to specialreg.hTheo de Raadt
2004-02-02PSL_XCRYPT is 0x40000000Theo de Raadt
2004-02-01Enable FXSR all the time, and cope with NPX/FXSR conversions; from netbsd.Theo de Raadt
Pass SSE/SSE2/XCRYPT flags out via syctl, and prepare for being able to do xcrypt-* in userland; ok naddy, tested a lot by pvalchev and jolan, also works on amd64 in 32bit mode
2004-02-01Sync user ldt code with NetBSD:Alexander Yurchenko
- finally remove it from pcb, it's a pmap thing only - more sanity checks - better lockin - may be something else Fixes panics when using apps requiring it (mplayer-win32 e.g.). Problem found and test espie@. OKs from miod@ (sshhh, don't tell anyone) and art@.
2004-01-31rename SIMD/SIMD2 to SSE/SSE2Theo de Raadt
2004-01-28sniff. first file i add'd is first to be rm'dTed Unangst
2004-01-14Revert previous; this will cause issues on some machines, says drahn@Miod Vallat
2004-01-13Fix KERNTEXTOFF value (long) after the move to ELF.Miod Vallat
tricked-to-commit-into-i386-land millert@
2004-01-06adjust pentium_mhz when cpu speed changes. ok deraadtTed Unangst
2004-01-03put an mi wrapper around stdarg.h/varargs.h. gcc3 moved stdarg/varargs macrosMarc Espie
to built-ins, so eventually we will have one version of these files. Special adjustments for the kernel to cope: machine/stdarg.h -> sys/stdarg.h and machine/ansi.h needs to have a _BSD_VA_LIST_ for syslog* prototypes. okay millert@, drahn@, miod@.
2004-01-01Crank SHMMAXPGS from 2048 to 8192. Other platforms where peopleTodd C. Miller
are using lots of shm may also wish to also increase SHMMAXPGS. OK deraadt@
2003-12-29Properly recognize SiS CPU family;Alexander Yurchenko
tested by Ian Zagorskih <ianzag@megasignal.com>. ok mickey@
2003-12-20add pentium 4 thermal control circuit (tcc) driver. can use this toTed Unangst
limit power consumption with the hw.setperf sysctl. tested by beck and grange
2003-12-19add cpu_ecxfeature to cpu.h, and stop locally externing it and cpu_feature.Ted Unangst
ok deraadt@
2003-12-18add new hw sysctls, cpuspeed and setperf to control cpu frequency.Ted Unangst
convert longrun support to use new sysctls. add enhanced speedstep support, based on code by Michael Eriksson. idea, help testing & ok deraadt@
2003-11-27This var is volatile. gcc3 complains if the CV qualifiers don't match.Marc Espie
ok toby, mickey
2003-11-16convert __attribute__((__packed__)) to __packed so that parsers unawareAnil Madhavapeddy
of gcc extensions have more of a chance. ok mcbride@, no objections from millert@, deraadt@
2003-10-28guard cpu register functions from userland, otherwise they appear whenAnil Madhavapeddy
including stuff like <sys/sysctl.h> deraadt@ ok
2003-10-24No need to have several implementations of {read,write}_eflags(),Alexander Yurchenko
put it to cpufunc.h. ok weingart@ tedu@
2003-09-23Replace select backends with poll backends. selscan() and pollscan()Todd C. Miller
now call the poll backend. With this change we implement greater poll(2) functionality instead of emulating it via the select backend. Adapted from NetBSD and including some changes from FreeBSD. Tested by many, deraadt@ OK
2003-09-08match size of fpreg w/ the save87 structs as figured in discussion w/ jason ↵Michael Shalayeff
two months ago and the rediscovered by grange@ and then probably by some more other people in mongolia who had no inet to report it unfortunately; this also fixes problem w/ catching sigsegv after recent bcopy(,,sizeof) fixes; deraadt@ ok
2003-08-21fix symbol name error in mcount functionality in a.out->ELF transition.Dale Rahn
This changes the real definition to __mcount, but with a weak mcount for compat. On the next major bump the weak alias should be removed. Without this diff the compiler symbol mcount conflicts with the user mcount() function.
2003-07-28nuke more MATH_EMULATEJason Wright
2003-07-25s/CPUID_SYS2/CPUID_SEP/ to avoid confusionMichael Shalayeff
2003-07-25make sure SYSENTER registers are 0 at boot time. should fix an unusualTed Unangst
DOS reported by Michal Zalewski to bugtraq. ok mickey@ toby@
2003-07-07function for cpuid instruction. pulled from longrun into generic code.Ted Unangst
ok deraadt mickey