summaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2017-07-23Prevent a possible "MAC context already removed" panic in iwm_auth().Stefan Sperling
This panic happens if iwm_newstate_task() sleeps and wakes up after iwm_stop(). The error path of iwm_newstate_task() attempts to reset driver flags and firmware state, which should not be done if iwm_stop() has already done so. A more comprehensive fix for such races is being worked on but not ready yet. For now, check the generation counter in the error path of iwm_auth() and only clean up state if that counter hasn't changed. Problem reported and workaround tested by giovanni@
2017-07-23use ssize_t; ok jcaTheo de Raadt
2017-07-23Time comparison variables should be time_t (I did check for signed vsTheo de Raadt
unsigned handling) ok jca
2017-07-23Cease pretending to support option 33 (classFULL static routes). TheyKenneth R Westerback
are dead to the internet. And the current code actually creates /32 routes since kernel magic for classfull routes died a while ago. ok phessler@ claudio@ reyk@ sthen@
2017-07-23Cleanup apm:anton
- Zap trailing whitespace in license, noted by tb@ - Sort and align function prototypes - Add static and __dead when appropriate in order to compile using `clang -Wall` - Zap variable names from function prototypes, also noted by tb@ - Break long lines ok tb@
2017-07-23recent perls have more fun rules related to _ and -Marc Espie
2017-07-23Add RK3288 support.Mark Kettenis
2017-07-23Replace CPWAIT with an isb instruction, which is the proper way to make sureMark Kettenis
CP15 updates are visible. Also add an isb instruction before switching on the MMU to make sure that all the MMU-related CP15 registers updates are visible. Makes booting on a Cortex-A12/A17 get a bit further. ok drahn@, jsg@
2017-07-23Recognize '-weak' in our libtool, helps at least xine-lib to build.Vadim Zhukov
okay aja@
2017-07-23Fix SPLICE in LT::UList, allowing more ports to build.Vadim Zhukov
Initial prodding and bulk builds by aja@.
2017-07-23from libc malloc: only access offset if canaries are enabled *and* size > 0Otto Moerbeek
and initialize ->offset to a definite value in the size == 0 case
2017-07-23Splitting early startup code out of locore.S into locore0.S, like otherKenji Aoyama
architectures. "Looking good" deraadt@
2017-07-22acpibat: look for _BIX first before falling back on _BIFJoshua Stein
ACPI 4.0 deprecated _BIF for battery status so some newer machines are exporting _BIX which extends _BIF's package with a few extra fields. When using _BIX, export the new discharge cycle count as a sensor. feedback from kettenis and anton ok deraadt
2017-07-22No need for sys/ioctl.h any moreJeremie Courreges-Anglas
2017-07-22Rename variables for clarity.Visa Hankala
It does not seem necessary to provide initial value for `sicode'. When `signal' gets set, `sicode' is set as well. OK deraadt@, kettenis@
2017-07-22Initialize the brightness to a valid value if found.anton
ok deraadt@
2017-07-22Add set_routes() and move bind_lease() route magicKenneth R Westerback
into it. Swap parameter order in add_default_route() to be consistant with other add_*_route() functions.
2017-07-22Delete extra whitespacesJeremie Courreges-Anglas
2017-07-22err(1, "whatever") -> err(1, NULL) for malloc(3) failuresJeremie Courreges-Anglas
2017-07-22use automatic variables rather than globals; no functional change;Ingo Schwarze
from ScottCheloha at gmail dot com on tech, tweaked by me
2017-07-22Make the kernel panic if an invalid state transition occurs in net80211.Stefan Sperling
Triggers on driver bugs such as those which were fixed in rsu(4) recently. ok kevlo@
2017-07-22Add frame length range checks to the input path of iwm(4).Stefan Sperling
No security benefit since the firmware has DMA access but we should not have such bad examples in our source tree. ok kevlo@
2017-07-22Fix length checks in EAPOL key frame parsing.Stefan Sperling
Problem reported by Ilja Van Sprundel. ok tb@ kevlo@
2017-07-22Bring SIGBUS and SIGSEGV handling better in line with the otherVisa Hankala
architectures. OK kettenis@, deraadt@
2017-07-22no need to cast signal(3) and fprintf(3) return values to (void)Ingo Schwarze
in standard use cases with constant arguments, and prefer getprogname(3) over __progname; from ScottCheloha at gmail dot com on tech
2017-07-22Tolerate imprecise fault address on sparc64 where the hardware doesn't provideMark Kettenis
tho low order bits. ok deraadt@, visa@
2017-07-22Prefer the definition of a variable in the executable over a definitionMark Kettenis
in a shared library, even for commons. This is what the current generation of linkers (current bfd, gold, lld) do. This fixes the relocation R_X86_64_PC32 against `xxx' can not be used when making a shared object; recompile with -fPIC warnings that we have seen in ports with clang. This change is somewhat suspect as a similar change was reverted in upstream binutils at some point. However that was for another corner case on an architecture (s390) that we do not run on. Tested by naddy@, sthen@
2017-07-22Whitespace.Kenneth R Westerback
2017-07-22Oops. Missed chunk keeping /32 netmask magic.Kenneth R Westerback
2017-07-22Fix the various SIGSEGV and SIGBUS scenarios to match what we do onMark Kettenis
amd64/arm64/armv7/i386/sparc64. ok visa@
2017-07-22Fix the various SIGSEGV and SIGBUS scenarios to match what we do onMark Kettenis
amd64/arm64/armv7/i386. ok visa@, deraadt@
2017-07-22Rename *_add_address to *_set_address to reflect that theKenneth R Westerback
lease address replaces all current addresses. Shuffle some code from bind_lease() into set_mtu()/set_address() to shrink bind_lease().
2017-07-22Introduce jiffies, a volatile unsigned long version of our ticks variableMark Kettenis
for use by the linux compatibility APIs in drm(4). While I hate infecting code in sys/kern with this, untangling all the of having different types and different signedness is too much for me right now. The best strategy may be to change ticks itself to be long but that needs some careful auditing. ok deraadt@
2017-07-22Add missing RCS IDs and zap redundant SRCS from Makefile.anton
2017-07-22Favor err() over perror() in example.anton
ok schwarze@
2017-07-22Cleanup previous release patches and backup kernel _before_ applying newAntoine Jacoutot
patches. This will mitigate possible /var overflow (previous behavior was to clean up _after_ applying all patches which means we would have rollback patches from the previous and current release before the previous ones got cleaned).
2017-07-22Use monotonic clock for the time command in csh and ksh.anton
From Scott Cheloha ok tb@
2017-07-22zap trailing whitespace;Jason McIntyre
2017-07-22rework the page a bit, clarify a few things, maybe better wordingTed Unangst
2017-07-21Use the actual width written rather than the possible width to clear.Nicholas Marriott
2017-07-21Stress the fact that individual pvbus device nodes correspond toMike Belopuhov
a single key-value store and if there are other ones available, different nodes must be used to access them.
2017-07-21Add fields for wsmouse/touchpad configuration.Ulf Brosziewski
2017-07-21Rename _hn to AI_HOSTNAME and export it as global variable so it'sRobert Peichaer
visible in the MODE.site scripts too. This is useful for reporting back the autoinstall status at the end of installs or upgrades from within the MODE.site scripts. OK krw@ tb@
2017-07-21Align ifstart() in netstart and install.sub.Robert Peichaer
- in netstart, rename _file to _hn referencing hostname.if files - in install.sub switch ifstart() to be used with _if instead of _hn as parameter ok krw@ tb@
2017-07-21Do not permit USB ioctl handler malloc(9)'s to block as this interferesians
with expected behavior. OK mpi@ patrick@
2017-07-21Show Synaptics Model IDs.Ulf Brosziewski
2017-07-21Replace MD _bus_dmamap_* function calls with MI onesMike Belopuhov
2017-07-21syncMike Belopuhov
2017-07-21Multiple virtualization layers may be available at the same timeMike Belopuhov
reachable through different pvbus device nodes. Suggestion and OK deraadt, OK reyk
2017-07-21s/set_interface_mtu/set_mtu/gKenneth R Westerback