summaryrefslogtreecommitdiff
path: root/sys
AgeCommit message (Collapse)Author
1998-05-22Start the timeout's if sparc_led_blink is initialized to trueJason Wright
1998-05-21typoMatthieu Herrb
1998-05-20Increase NVNODE.Jason Downs
1998-05-20syncJason Downs
1998-05-20I think this is more correct than the last; it makes more sense, at least.Jason Downs
1998-05-20magma serial driver; plunky@skate.demon.co.ukTheo de Raadt
1998-05-20syncJason Downs
1998-05-20VIA Apollo VP3 AGPsetJason Downs
1998-05-20Enable NTP again now that mickey has solved the pentium timer problem.gene
1998-05-20fix strtol on one character long numbersArtur Grabowski
1998-05-19cross-compile readyMichael Shalayeff
1998-05-19Wall for non-IPSEC caseTheo de Raadt
1998-05-18start using new apiver thingsMichael Shalayeff
1998-05-18define bitmapped bootarg api versionMichael Shalayeff
1998-05-18add more signals to setuid part of cansignal()Theo de Raadt
1998-05-18first step to the setsockopt/getsockopt interface as described inNiels Provos
draft-mcdonald-simple-ipsec-api, kernel notifies (EMT_REQUESTSA) signal userland key management applications when security services are requested. this is only for outgoing connections at the moment, incoming packets are not yet checked against the selected socket policy.
1998-05-18ok, i steal this commit from niklas, just to make vax consistant w/ other ↵Michael Shalayeff
archs (;
1998-05-18The warnings shouldn't tell that this is ffs. It's confusing.Artur Grabowski
1998-05-18new cpureg.h from NetBSD plus modifications to use the changed macro namesTodd C. Miller
1998-05-18newer asc driver from NetBSDTodd C. Miller
1998-05-17bus.h from NetBSD, does not implement OpenBSD extensions yet.Todd C. Miller
1998-05-17convert CANSIGNAL() macro into a function, and then add significant constraintsTheo de Raadt
regarding setuid/setgid targets (this is permitted by POSIX, and solves a number of security issues). This is due to a conversation months ago with dholland@hcs.harvard.edu.
1998-05-17fix tdb_delete() when using SPI chains.Niels Provos
1998-05-17for acct(), vn_open() with O_APPEND flag so that append-only files work;Theo de Raadt
problem pointed out by andrew@untraceable.net
1998-05-17Intercept all targets in cross-modeNiklas Hallqvist
1998-05-17Automatically detect cross-compilation case. If you have an i386Niklas Hallqvist
(well that is the only tested host environment for the time being) and would like to test the support, do this: cd /usr/src make TARGET=vax cd sys/arch/vax/conf config MYKERNEL cd ../compile/MYKERNEL make and you should get a nicely compiled vax kernel. This is WIP, so do not expect robustness quite yet.
1998-05-17define DISKLABELDEV(dev) macroMichael Shalayeff
1998-05-16shorten device probe messages to 2 linesTheo de Raadt
1998-05-16syncimp
1998-05-16Add DeskStation Technology and Evolution RISC PCimp
1998-05-16syncTheo de Raadt
1998-05-16new compaq devsTheo de Raadt
1998-05-15syncTheo de Raadt
1998-05-15+ opti RM861HATheo de Raadt
1998-05-14change debug poke colorsMichael Shalayeff
1998-05-14nicer boot messageMichael Shalayeff
1998-05-14We are OpenBSD. That and international :-)Niklas Hallqvist
1998-05-14KNFNiklas Hallqvist
1998-05-14* Better support for XR16C850.Jason Downs
* Copyright.
1998-05-14* Better XR16C850 support.Jason Downs
1998-05-14* Better XR16C850 support.Jason Downs
* Copyright. * General cleanup.
1998-05-13more careful setupTheo de Raadt
1998-05-13making kernel, drinking LI tea and writing an opus magnum enspiredMichael Shalayeff
by the freebsd commit logs: first of all i should state that we are the only os that uses pentium counters in their microtime (freebsd, netbsd, etc checked). the bug fixed derives from the fact that #UD (divide by zero) is generated in two cases: a) divide by zero; b) divide overflow. so, the later case was the one causing panics on reboot. due to the divl insn design in the intel processors, the possible sizes of operands and results are: 16/8=8,8; 32/16=16,16; 64/32=32,32. the division will generate an #UD exception when the quotient won't fit into result, that is 32bit in our case (64bit is divided on 32bit). problem solved by normalizing the dividend by the divisor, so the quotient would not result a divide overflow. tested by: mickey, gene, mickey. talked over: mickey, gene, marc, maillists, mickey.
1998-05-13bad toddTheo de Raadt
1998-05-13don't printf in probe routines.Niels Provos
1998-05-13Use DEV_BSIZE instead of the wasteful MAXBSIZE for buffer size.Niklas Hallqvist
1998-05-13main returns int, not voidTodd C. Miller
1998-05-12typoTodd C. Miller
1998-05-11s/ffs_mountroot/dk_mountroot/, name RD disks "hd" to not conflictNiklas Hallqvist
with the MI ramdisk device
1998-05-11Do not allow mapping of immutable/append-only in a way that allows writing.Niklas Hallqvist
Information about the problem gathered from NetBSD, but solved in a different albeit similar way.
generated by cgit v1.2.3 (git 2.25.1) at 2024-12-14 09:12:52 +0000