summaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2019-02-17sprinle some casts to quiet compiler; ok millertTheo de Raadt
2019-02-17Make bridge_rtupdate() return an error value instead of a pointer.Martin Pieuchot
2019-02-17Publish new attached device once it is fully initialized.Martin Pieuchot
Should prevent a race seen by florian@ where userland emits an USB_DEVICEINFO ioctl(2) for a device whose product/vendor names haven't yet been filled. ok florian@
2019-02-17Convert openssl(1) pkeyutl to the newer style of option handling.Kinichiro Inoguchi
ok jsing@ tb@
2019-02-17On log verbosity change we need to restart all resolvers: add missingFlorian Obser
DoT resolver.
2019-02-17Pass verbosity changes to captive portal process.Florian Obser
2019-02-17s/unwind_/uw_/ to save screen real estateFlorian Obser
2019-02-17 s/unwind_/uw_/ to save screen real estate; fix style(9) issues while hereFlorian Obser
2019-02-17Has to be SOCK_STREAM obviouslyTheo de Raadt
2019-02-16use MINIMUM() instead of MIN() as elsewhere in the treeTheo de Raadt
2019-02-16Sprinkle a few ifdefs for _LOCORE and _KERNEL and reorder a few linesPatrick Wildt
so that pmap.h can be included as part of the mmap_hint regression test. From Moritz Buhl ok bluhm@
2019-02-16Do not look at next key byte if the length is 0, originally from ShingoNicholas Marriott
NISHIOKA in GitHub issue 1601.
2019-02-16SSID/LLADDR/LINK_STATE_IS_UP changes must discard the previousKenneth R Westerback
proposal to ensure the new lease is applied to the interface.
2019-02-16nasty ugly invisible spacesTheo de Raadt
2019-02-16Document some long optionsTheo de Raadt
2019-02-16--archive is same as -aTheo de Raadt
2019-02-16document -a, -D, --devices and --specialsFlorian Obser
2019-02-16add --help optionTheo de Raadt
2019-02-16document --versionTheo de Raadt
2019-02-16add missing -no-OPT long optionsTheo de Raadt
ok florian
2019-02-16add fsync to STANDARDS;Jason McIntyre
2019-02-16sync with kristaps, commit 42166a639caa877efd91e73e288b89abf6b7cfd3Florian Obser
Remove spaces at end of line.
2019-02-16sync with kristaps, commit 1842d31d600f2834ae78fe0d99a29519a853d75cFlorian Obser
Make sender mostly nonblocking for writes. This takes a lot of logic once in blocks.c and puts it directly into sender.c. It allows running openrsync against itself without deadlocks.
2019-02-16sync with kristaps, commit d01f1b87a06c25353de28a9df45b3e202614b336Florian Obser
Add the lowbuffer routines. I'll use this in subsequent commits, but they pre-add the multiplexing frame instead of doing so during write. While here, clean up a lot of ERRX -> ERRX1 and add documentation.
2019-02-16sync with kristaps, commit d7c4fb8ac88845aa08900d5d0ec469257f63a339Florian Obser
Use a static single block instead of multiple writes.
2019-02-16sync with kristaps, commit f5c1e234990877cd274b7bcdd4d9883104cd1568Florian Obser
Make sender have reentrant sending and prioritised reading. This way, the sender gets data off the wire as quickly as possible. While here, remove dead wood from blocks.c (blk_merge).
2019-02-16sync with kristaps, commit ad6f9d84aa041d28b93cdf885a583a09f58b1b88Florian Obser
Remove superfluous includes.
2019-02-16sync with kristaps, commit 9b79b4a3d06c810304321d5b58544751b5d9fefdFlorian Obser
Fast-track reads back into a read loop to avoid the buffer with writes while there are still reads pending. This resolve some bottlenecking.
2019-02-16use service "rsync" rather than "873"Theo de Raadt
2019-02-16getaddrinfo doesn't imply "DNS", borrow terminology from sshTheo de Raadt
ok benno
2019-02-16-D bits, from kristapsFlorian Obser
2019-02-16We need to send/receive the rdev after gid and before symlink.Florian Obser
From kristaps, thanks!
2019-02-16When specifying the time in the form HHMM or HH:MM, if that timeChristian Weisgerber
is already past, the next day is assumed. Handle the case where we're still in the same hour but the minutes are past. ok deraadt@ millert@
2019-02-16Use starting client cwd in config file, GitHub issue 1606.Nicholas Marriott
2019-02-16Implement the conv=fsync feature which does an fsync(2) after theAlexander Bluhm
final write to output. GNU dd also has this. It is useful for write performance measurement or guaranteed writes to reliable storage. OK kn@ tedu@ deraadt@
2019-02-16I did some work here.Florian Obser
2019-02-16implement -D and in turn -aFlorian Obser
"you snooze, you get collisions" deraadt@
2019-02-16Handle file permissions in rsync_set_metadata().Florian Obser
We set permissions either when we create a new file or when the -p flag is set. "you snooze, you get collisions" deraadt@
2019-02-16Introduce mkstempat(), a variation on mkstemp(3) and mkstemplinkat().Florian Obser
mkstempat() works exactly like mkstemp(3) except that it replaces open(2) with openat(2) so that it can be used in rsync_downloader() to easily deal with relative paths. mkstemplinkat() works somewhat like mkdtemp() to create a template symlink. Use the mkstemplinkat() to create or update symlinks and overwrite existing objects including empty directories that might exist under the destination name. "you snooze, you get collisions" deraadt@
2019-02-16Add chown pledge so that root can gift files to other uids.Florian Obser
"you snooze, you get collisions" deraadt@
2019-02-16a minimal --version long-option, so someone can unconfuse themselvesTheo de Raadt
2019-02-16use gettimeofday + TIMEVAL_TO_TIMESPEC to create timespec with betterTheo de Raadt
than 1 sec resolution ok benno
2019-02-16If anyone ports this to 6.4 they'll just comment out unveil, don't wasteTheo de Raadt
lines telling them.
2019-02-16unwrap lines which do not need wrappingTheo de Raadt
2019-02-16Add tests for pkeyutlKinichiro Inoguchi
2019-02-16vmm(4): report PA bits in CPUID[0x80000008]Mike Larkin
We were previously hiding this information from the guest but there really isn't a reason to do that. Hiding this value limits the PA size to 36 bits (64GB), which I discovered the hard way when creating some big VMs recently.
2019-02-15Remove KASSERT() for now. It triggers when destroying lo(4) of a rdomainClaudio Jeker
because the rtable_l2 is modified before calling rt_ifa_del. Triggered by regress test and reported by Moritz Buhl mbuhl at mbuhl dot me
2019-02-15zap trailing empty lineanton
2019-02-15The underlying storage for builtin ddb variables are of type int butanton
referenced to using a pointer to long. When writing to such a variable, cast it to the correct type. Writing would otherwise on 64-bit architectures cause the next variable adjacent in memory to also be modified. ok deraadt@ visa@
2019-02-15With an opportunistic check performed at every trap, we insist userlandTheo de Raadt
sp must be on a MAP_STACK page. Relax the check a bit -- the sp may be on a PROT_NONE page. Can't see how an attacker can leverage that situation. (New perl build process contains a "how many call frames can my stack hold" checker, and this triggers via the MAP_STACK fault rather than the normal access check. The MAP_STACK check still has a kernel printf as we hunt for applications which map stacks poorly. Interestingly the perl code has a knob to disable similar printing alerts on Windows, which apparently has a feature somewhat like MAP_STACK!) ok tedu guenther kettenis
generated by cgit v1.2.3 (git 2.25.1) at 2025-01-15 12:51:15 +0000