summaryrefslogtreecommitdiff
path: root/sys/arch
AgeCommit message (Collapse)Author
2011-11-09Fix manual wfi instruction encoding.Miod Vallat
2011-11-09Missing argument to error printf()Miod Vallat
2011-11-09In pmap_remove(), don't forget to do tlb flushes even in the !pmap_cachevivtMiod Vallat
case. ok drahn@
2011-11-09Use the new priority as the priority to check against,Dale Rahn
not the current (not yet lowered) priority. problem spotted by miod.
2011-11-09External ampintc interrupts need to add 32 to the interrupt number,Dale Rahn
while 'internal' should not. no complaints uwe@ miod@
2011-11-08Cookies? Beagle don't need no stinkin' cookies. To get boot paramsKenneth R Westerback
anyways. ok (sight unseen) deraadt@ drahn@
2011-11-08Make the userland bits actually compile.Mark Kettenis
ok guenther@
2011-11-08Let beagle GENERIC work as 'swap generic'.Kenneth R Westerback
ok deraadt@ drahn@
2011-11-08consistant "config bsd" lines; ok miodTheo de Raadt
2011-11-08Garbage collect now unused MKDEP definitions. ok deraadt@Matthieu Herrb
2011-11-08merge cpu_coredump() back to where it belongs, ok drahn miodTheo de Raadt
2011-11-08merge cpu_coredump() back to where it belongs; ok miod drahnTheo de Raadt
2011-11-08remove support for big endian; ok drahnTheo de Raadt
2011-11-08compat_13 support died ages agoTheo de Raadt
2011-11-08Save and restore cr27 and add the necessary glue to use it as the userlandMark Kettenis
thread register. ok guenther@, jsing@
2011-11-08On PCXS processors, reading %cr26 and %cr27 from userland causes aMiod Vallat
privileged register trap. Cope with this in the trap handler to let userland use these registers; this will be necessary in the not-so-distant future.
2011-11-08arm has no need for procfsTheo de Raadt
2011-11-08procfs for solaris compat? not anymore; ok miodTheo de Raadt
2011-11-07Cleanup and make initialization more readable. ok drahn@ uwe@Miod Vallat
2011-11-07TLS-lite support for i386Philip Guenthe
2011-11-07Lift the copyout() from i386_get_threadbase() to sys_sysarch(), so thatPhilip Guenthe
it can be used correctly from linux_sys_get_thread_area() and tcb.h. linux_sys_set_thread_area() should use the copied in value instead of deferencing SCARG() again. i386_set_threadbase() should only update curcpu's GDT if it's being called for curproc. ok pirofti@
2011-11-07Fix the definition of L1_S_V7_AP().Miod Vallat
2011-11-06For now turn back on the pmap PTE modification in pmap_l2ptp_ctorDale Rahn
as was there before the PMAP_INCLUDE_PTE_SYNC changes. Beagle xM is happier this way.
2011-11-06Correctly report WB and WT cache modes on v7 (they were swapped). ok uwe@Miod Vallat
2011-11-06Build boot(8) for i386 in the same manner that we build boot(8) for amd64 -Joel Sing
specify a list of source files to compile, instead of building and linking against libraries. ok deraadt@
2011-11-06Fix GTIMER_STATUS offsetMiod Vallat
2011-11-06Further on the road to working.Dale Rahn
2011-11-05More effort to use filesystem time if the rtc doesn't exist, thanks miod.Dale Rahn
2011-11-05Introduce clock registration code, to deal with having multipleDale Rahn
delay/cpu_initclock operations from different drivers.
2011-11-05Remove incorrect logic leading to skip page table flushes becauseMiod Vallat
PMAP_NEEDS_PTE_SYNC would get defined to zero on kernels lacking StrongArm support. discussed with and ok drahn@
2011-11-05rename gptimer_delay to gptimer_wait, as it waits on the bit to be clearedDale Rahn
dont include bus.h twice.
2011-11-05Many fixes.Dale Rahn
2011-11-05Add a NOMAN= stanzaMiod Vallat
2011-11-05Attempt to configure the interrupt controller and global timer on cortex-a9Dale Rahn
2011-11-05Initial drivers for Cortex A9 interrupt controller and global timer as clock.Dale Rahn
2011-11-02display AMD/extended (0x80000001) cpuid flags and remove theJonathan Gray
Cyrix 3DNOW flag from normal cpuid flags as it will show up in extended flags. ok kettenis@
2011-11-01Drop support for the HP MMU. It was only found on two hp300 systems (modelsMiod Vallat
320 and 350), which have been unsupported since a bunch of release already, because this annoying MMU does not have TT registers (or if it does, it is not documented) and thus went in the way of PMAP_DIRECT for hp300. In other words: remove a bunch of code which was either #ifdef'ed out or had no chance to run in real life.
2011-11-01Rework disklabel handling to correctly support label spoofing, and thusMiod Vallat
fix DIOCGPDINFO ioctl behaviour in the process. Tested in SIMH only due to the lack of MSCP disks, RX02 floppies untested but shouldn't be affected by this diff.
2011-11-01Remove orphaned comments.Miod Vallat
2011-10-29Make sure we don't accidentally use ISA or EISA interrupt mappings on PCIMark Kettenis
busses. tested by krw@
2011-10-28Enable interrupts while processing off level interrupts. simplify codeDale Rahn
instead of having two levels of looping. commit it for testing deraadt@
2011-10-28Ignore devices with irq 0 on mpcpci bus, the only devices seen thisDale Rahn
way is hardware not found in the openfirmware tree. diff from kettenis@ commit it deraadt@
2011-10-27Specify the TLS variant for each platform.Philip Guenthe
2011-10-26Make this compile.Miod Vallat
2011-10-25Replace the naive 88110 tlb update code, which would always invalidate theMiod Vallat
whole tlb (32 of 'em), with smarter `tlb probe and update with new pte if tlb match found' code. This makes the 88110-specific pmap_update() unnecessary, as updates are no longer aggregated to avoid the number of flushes. This also makes tlb handling similar between 88100 and 88110, from the pmap's point of view, so there is no need to use different routines. No impact on 88100, no user-noticeable performance change on 88100 GENERIC, slight improvement on 88110 GENERIC.MP.
2011-10-25Bring luna88k disklabel handling code in sync with sparc, except for theMiod Vallat
SunOS vs UniOS differences for the legacy part (first eight partitions), which are preserved. Tested by aoyama@
2011-10-24Introduce a pluggable interrupt controller infrastructure for beagle,Dale Rahn
to allow panda to share the port.
2011-10-24The cgtwo control registers only support 16-bit accesses, and respond withMiod Vallat
a bus error to larger widths. Declare all control register bitfield structs with uint16_t instead of u_int, to force gcc4 to use short accesses.
2011-10-24Revert last change, reported increased stablility with the change wasDale Rahn
grossly overstated.
2011-10-23Remove comment; the question can be answerd affirmative. Fix a coding styleMark Kettenis
nit while I'm there.
generated by cgit v1.2.3 (git 2.25.1) at 2024-12-04 11:27:38 +0000