summaryrefslogtreecommitdiff
path: root/sys/arch
AgeCommit message (Collapse)Author
2007-11-16fix the bus_space #define nightmare, so that amd64 and i386 are much moreTheo de Raadt
uniform. as a result shared code like acpi needs less #ifdef's ok marco kettenis
2007-11-16Remove microtime calls and RNG benchmarking from the pchb attachmentMike Belopuhov
code. This is due to the fact that we do timecounter initialization after autoconf(9). ok deraadt
2007-11-16something has to go, unfortunatelyTheo de Raadt
2007-11-16do not make up a device name for interrupts.. give it the dv_xnameTheo de Raadt
2007-11-15enable most acpi functionality by default. now instead of 'enable acpi',Theo de Raadt
you use 'disable acpi' or 'disable apm' withn you encounter problems (depending on which kind of problems). if we work hard enough, this can remain the situation by the time we ship the next release. otherwise, we will re-disable acpi... so let's crackin'
2007-11-15remove the #ifdef ACPI_ENABLE. the EC codepath is either safe enough (andTheo de Raadt
will be improved), or it is unsafe because AML assumes it must be there ok almost everyone
2007-11-15Check for curcpu()->ci_softintr being nonzero before invoking dosoftint(),Miod Vallat
in the common case, this saves us two potentially expensive setipl() calls.
2007-11-15Make a few end-of-trap-handling symbols global, to get better ddb traces.Miod Vallat
Also, pick better symbol names in the sir/ast processing.
2007-11-15At last, port the mvme88k eh.S r1.66 fixes to aviion.Miod Vallat
2007-11-15Resolve the struct trapframe tf_mode/tf_flags duality by getting rid ofMiod Vallat
the tf_mode name. This field only contains flags used internally by the exception processing code.
2007-11-15Stop referring the initial kernel stack as the ``interrupt stack''. It'sMiod Vallat
been years since it has last been used for that purpose, so name it the initialization/startup stack. While there, do not store the initialization stack in cpu_info, and have secondary_pre_main() return its value so that the bootstrap code does not need to fetch it from cpu_info. This might be reconsidered when the startup stacks will be freed after they are not used anymore, but there are more things to do first.
2007-11-15Attach to "rsc-console". Hopefuly this will give us a proper console onMark Kettenis
the Sun Fire 280R/V480/V880 when the console is directed to the RSC card.
2007-11-15in MP kernels, use smbios to count number of cpus. if > 1 disable apm, soTheo de Raadt
that acpi can get a chance at controlling the machine. done with marco, and even tested on the nastiest case: smbios <2.5 w/ HTT
2007-11-15Split CPUID into 2 32 bit fields.Marco Peereboom
ok deraadt
2007-11-15``stastic'' is a fantastic typo.Miod Vallat
2007-11-15Clarify when smbios added multi core support and add the missing stringMarco Peereboom
designations. ok deraadt
2007-11-15oops, processor structs need to be __packed.Marco Peereboom
prompted by deraadt
2007-11-15Add struct for processors.Marco Peereboom
ok deraadt
2007-11-14Cache curcpu() value into a local variable when it is used more than once inMiod Vallat
a function, so that it does not get reloaded from cr17 every time.
2007-11-14No need to check whether we are the primary processor in m188_{clock,stat}intr,Miod Vallat
since the corresponding interrupt source is enabled on the primary processor only.
2007-11-14Merge the ci_alive and ci_primary boolean values of struct cpu_info intoMiod Vallat
a single ci_flags bitfield. Also, set_cpu_number() will no longer set CIF_PRIMARY on the primary processor, it's up to the initialization code to do this.
2007-11-14When processing a data access fault, keep the kernel lock while invokingMiod Vallat
data_access_emulation() to complete the interrupted pipeline operations, as data_access_emulation() can fault in turn.
2007-11-14When servicing an exception, do not enable interrupts if they were notMiod Vallat
enabled when the exception occured. This should not happen in practice, but better be safe than sorry.
2007-11-14prettify the RCS id (less whitespace)Thordur I. Bjornsson
2007-11-14Remove eeprom.h.Mark Kettenis
2007-11-14Turn the spl* macros into functions.Thordur I. Bjornsson
Shaves a few bytes of the kernel. No measurable performance loss. ok drahn@, kettenis@
2007-11-14do not process requests linked to unused sessions. (crypto_freesessionMarkus Friedl
might happen between enqueuing a crypto request and scheduling of the crypto thread); ok hshoexer
2007-11-14Let ``machine cpu #'' hop to the given cpu.Miod Vallat
2007-11-14Be consistent in db_cmd_loop_done declaration.Miod Vallat
2007-11-13Fix printf format string.Mark Kettenis
2007-11-13Remove bsd_openprom.h.Mark Kettenis
2007-11-13Expose mace_bus_dma_tag and extend the crimebus extent. This allows theJoel Sing
existing bus interfaces to be used to access the GBE hardware found in the SGI O2. ok miod@
2007-11-13Give wscons_machdep.c half a spanking. This allows wsdisplay to compile onJoel Sing
OpenBSD/sgi and prevents it from stealing the console. ok miod@
2007-11-12Restart clock before invoking hardclock() and statclock(), otherwise theyMiod Vallat
drift in MP kernels.
2007-11-12Use isa_intr_establish() to establish the SCI interrupt handler; this makesMark Kettenis
sure that interrupt overrides from the MADT are applied. ok marco@, toby@
2007-11-12Split handling of unmaskable vs maskable ipi to make the core simpler,Miod Vallat
no functional change.
2007-11-12/dev/openpromMark Kettenis
ok miod@, deraadt@
2007-11-12fix some problems with lazy fpu context handling:Todd C. Miller
put the "fninit" where it belongs logically deal with XMM exceptions From NetBSD (drochner)
2007-11-12Select the right GDT entry for interrupt handlers on i386 systems whenChristopher Pascoe
using ioapics, so that interrupt time is accounted correctly. This fixes the problem that 0% interrupt time is always shown when running MP. ok dlg@, suspect this is correct art@, just an oversight niklas@
2007-11-11Rework ipi handling to delay acking the maskable ipi interrupt, so that itMiod Vallat
gets handled like a real hardware interrupt (which it is supposed to mimic anyway).
2007-11-11Give more information in ``machine cpu'' under ddb.Miod Vallat
2007-11-11Replace next_tick() with simpler C code that I can actually understand.Mark Kettenis
2007-11-11Add spurious interrupt detection to macebus. This allows for the detectionJoel Sing
of interrupt storms and enables the user to "boot poweroff" from ddb, instead of pulling the powercord. Initial code heavily reworked by miod@ ok miod@
2007-11-11Use two software interrupt sources per processor for IPIs, instead of onlyMiod Vallat
one, so that we can have maskable and unmaskable IPIs. Make the clock ipis maskable, and masked at IPL_CLOCK and above. This allows us to get rid of the retrig hack in setipl().
2007-11-11In dma_cachectl(), flush unconditionnaly on all processors, regardless of theMiod Vallat
cpu bitmask of the pmap.
2007-11-10weel -> wellJoel Sing
2007-11-10Activate umbg(4) on macppc.Marc Balmer
ok jsg, dlg.
2007-11-10Activate umbg(4) in i386.Marc Balmer
ok dlg.
2007-11-10Remove a fair amount of duplicated code by making cpu_mp_startup callMark Kettenis
cpu_initialize.
2007-11-10For single-processor kernels, only match the CPU we're running on.Mark Kettenis
generated by cgit v1.2.3 (git 2.25.1) at 2024-12-21 17:29:16 +0000