summaryrefslogtreecommitdiff
path: root/sys/arch
AgeCommit message (Collapse)Author
2001-09-29dec_kn20aa_intr_disestablish() implementation from netbsd; art@ okMichael Shalayeff
2001-09-29Enable pckbd support and use symbolic #defines for ctb_term_typeTodd C. Miller
instead of hard-coding values. art@ OK
2001-09-29com at ebusArtur Grabowski
2001-09-29com at ebusArtur Grabowski
2001-09-29Glue for com at ebus. From NetBSD, hacked to fit our com.Artur Grabowski
2001-09-29using #define is wrong, use #optionTheo de Raadt
2001-09-28and mention XP1000 in the right placeTheo de Raadt
2001-09-28Hack to work around bootpaths like:Jason Wright
bootpath: /pci@1f,4000/ide@3,0/ata@0,0/cmdk@0,0 The val[0] of ata appears to be the channel number, and cmdk is disk (cmdk@channel,disk)
2001-09-28cold is now staticly initialized.Miod Vallat
2001-09-28Variable declarations in .h files without extern keywords suck.Miod Vallat
Especially when not all of these variables are used.
2001-09-28Typos of the day.Miod Vallat
2001-09-28More cleaning, some style changes.Miod Vallat
2001-09-28"sys/foo.h" -> <sys/foo.h>Miod Vallat
2001-09-28No need duplicating defines.Miod Vallat
2001-09-28proper values for long double type (from NetBSD)Jason Wright
2001-09-28proper values for ieee rounding modes (from NetBSD)Jason Wright
2001-09-28Almost half of this file was ifdef DEBUG. Everybody knows that debuggingArtur Grabowski
code is just there for debugging so it doesn't need to be pretty or really correct. In most cases it's just a bunch of printfs that only the author understands for two hours after writing them. Remove all that noise to make this file actually readable and as a bouns fix a bug where any user could crash the kernel (not really crash, but triggering ddb from userland is equivalent to crash in my world even if you can just press 'c').
2001-09-28Don't spam the console with userland alignemnt errors.Artur Grabowski
2001-09-28upgrade twe to untested, add ami to brokenMichael Shalayeff
2001-09-28zs serial now works on some systems. Does not work with modemsDale Rahn
on some newer system due to power control issues. Has carrier detect problems with incoming (getty) on most/all systems. Outgoing seems to work fine (unless modem power issues prevent).
2001-09-28Commit a ugly, widespread change.Dale Rahn
Problem: no MI soft interrupts. Bad solution: use old hack that was never fully implemented to hook in zs soft interrupts. This allows zs to work almost reasonably. Still has issues with getty (carrier detect) and modems on newer machines (power control). Post 3.0 this must be cleaned up with real soft interrupts.
2001-09-28remove dups w/ conf.hMichael Shalayeff
2001-09-28remove dups w/ conf.hMichael Shalayeff
2001-09-28move cdev_decl(pf) into it's proper place, kill cdev_decl(ipl) at the same timeMichael Shalayeff
2001-09-27Two workarounds for the zs serial driver.Dale Rahn
Enable Interrupts (recieve/status) initially. Code currently does not enable them elsewhere. Use a timeout to scedule the soft interrupt, since no MI soft interrupt scheme works yet. This will have issues with high receive rates.
2001-09-27psycho_bus_[a|b] are essentially the same function; use a helper to avoid typosJason Wright
attempt to extract the faulting address from the iommu in the newly defined psycho_bus_error() function
2001-09-27When zs is fully working, it should be tty00 for macppc.Dale Rahn
2001-09-27Huge merge with NetBSD. For full details look at their revision history ↵Jason Wright
(1.26 -> 1.83)... Highlights: o tag queue support o fas366 support (doesn't work yet on sparc, but does on sparc64) o tons more (thanks to hugh, fgsch and beck for testing)
2001-09-27used to corrupt some memory -- no moreMichael Shalayeff
2001-09-26zs driver, from different srcs, including netbsd/macppc and out mac68k/zs, etc.Michael Shalayeff
2001-09-26be more specific in psycho_ue case (partially based on NetBSD)Jason Wright
2001-09-26missed in previous merge with NetBSD:Jason Wright
handle dma boundaries better
2001-09-26grab the "compatible" prom property and use that for searching the deviceJason Wright
tree as well as the "name". This allows us to differentiate 'sd' and 'wd' pretty easily (and makes it less ambiguous).
2001-09-26A little bit of voodoo borrowed from the linux driver: set theJason Wright
intr_retry_timer a little higher.
2001-09-26Merge with NetBSD:Jason Wright
better handling of boundary conditions add iommu_extract debugging function
2001-09-26change *int64*_t types to long long; ok art/jasonTheo de Raadt
2001-09-26add siop (and scsibus at siop)... This allows us to boot on E250Jason Wright
Thanks to chris@ for testing.
2001-09-25Add a bunch of relocation types.Artur Grabowski
2001-09-25shh, no point in printing useless kernel messages.Dale Rahn
2001-09-24Just one relocation type (needed for ld.so) right now.Artur Grabowski
2001-09-24fix back the kbd, sorryMichael Shalayeff
2001-09-23document some registers; copied from comreg.hTodd C. Miller
2001-09-23For serial console, drop into ddb on <break> if ddb.console is nonzeroTodd C. Miller
2001-09-23Rename kdb_init() to db_machine_init().Miod Vallat
Fix prototypos for the early console routines. Use even more PMAP_NEW syntactic sugar. Remove some debugging stuff. Use symbolic values instead of magic numbers. Fix safepri value. Implement machdep.console_device sysctl.
2001-09-23Declare machdep.conssole_device sysctl.Miod Vallat
2001-09-23rename kdb_init() to db_machine_init().Miod Vallat
2001-09-23kernel_pmap -> pmap_kernel()Miod Vallat
2001-09-23Don't define _C_LABEL here.Miod Vallat
XXX this file should probably disappear. soon.
2001-09-23Don't redefine pmap_kernel() locally.Miod Vallat
2001-09-23Worst abuse of C, ever.Miod Vallat