OpenBSD src changes summary for 2016-10-25 ==========================================
libexec/ld.so share/man sys/arch/armv7/armv7 sys/arch/armv7/exynos sys/arch/armv7/omap sys/dev/acpi sys/dev/ic sys/dev/pci sys/dev/pv sys/dev/usb sys/net sys/netinet sys/netinet6 usr.bin/ssh usr.bin/tmux usr.sbin/makefs usr.sbin/tcpdump == libexec =========================================================== 01/05 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/libexec ld.so ~ malloc.c > bring canaries for > page sized objects from libc to ld.so; ok guenther@ > (otto@) == share ============================================================= 02/05 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/share man ~ man9/cpumem_get.9 ~ man9/counters_alloc.9 > start documenting allocating per cpu data at boot time, and adding > memory for new cpus after theyve attached. > some tweaks from jmc@ who will probably find more things wrong after this. > im putting some suggestions in from bluhm@ shortly (dlg@) ~ man9/cpumem_get.9 ~ man9/counters_alloc.9 > bluhm@ suggests that it is enough to say that the allocations will > be zeroed, and how they're zeroed is an implementation detail that > doesnt really help the reader. (dlg@) ~ man9/cpumem_get.9 ~ man9/counters_alloc.9 > bluhm@ suggested pointing out that cpumem_malloc_ncpus and > counters_alloc_ncpus preserve the contents of memory used by the > boot cpu. (dlg@) ~ man9/counters_alloc.9 > provide an example based on the mbuf code (dlg@) ~ man9/counters_alloc.9 ~ man9/cpumem_get.9 > minor tweaks; (jmc@) == sys =============================================================== 03/05 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/sys arch/armv7/armv7 ~ armv7_machdep.c ~ armv7_machdep.h ~ platform.c > Remove now unused disable_l2_if_needed interface. > ok kettenis@ patrick@ (jsg@) arch/armv7/exynos ~ exynos_machdep.c > Remove now unused disable_l2_if_needed interface. > ok kettenis@ patrick@ (jsg@) arch/armv7/omap ~ omap_machdep.c > Remove now unused disable_l2_if_needed interface. > ok kettenis@ patrick@ (jsg@) dev/acpi ~ acpi.c ~ acpiprt.c ~ bytgpio.c ~ chvgpio.c ~ dsdt.c ~ dsdt.h ~ dwiic.c ~ sdhc_acpi.c ~ tpm.c > Add the resource index to aml_parse_resource() callback function. > This is needed by an upcoming acpiec commit that handles machines breaking > the current ACPI specifications. > Change suggested by and ok kettenis@, guenther@ (pirofti@) ~ acpidev.h ~ acpiec.c > Deduplicate the CRS parsing code and handle broken ACPI implementations. > This change was prompted by a report from Ilya Kaliman where acpiec(4) > failed to initialize on his Acer Aspire S7. > Further investigations showed that the machine had 4 instead of 2 resources > defined inside the EC device which is a violation of all existing ACPI > revisions. > We now handle such machines by using the aml resource parser directly, > which > lead to the removal of the existing one from acpiec.c, and ignoring extra > resources (although they are reported in the dmesg). > In the process I also done a couple of more style(9) fixes. > Suggested by and ok guenther@ (pirofti@) dev/ic ~ nvme.c ~ nvmevar.h > mask and unmask the interrupt source in an intx specific intr handler. > it seems devices using levelled intx interrupts need to explicitely ack > interrupts by masking and unmasking the source around the completion > ring handling. without this completions can be lost, which in turn > causes long (permanent?) stalls in the block layer under heavy write > load. > ive experienced this problem with an intel nvme part that only has > intx and msix support. because we dont support msix yet we only > use intx on it. it appeared to lock up before this fix. > this has been tested on both that intel board and a samsung with msi. > this fix was based on work found in code by nonaka (dlg@) dev/pci ~ nvme_pci.c > mask and unmask the interrupt source in an intx specific intr handler. > it seems devices using levelled intx interrupts need to explicitely ack > interrupts by masking and unmasking the source around the completion > ring handling. without this completions can be lost, which in turn > causes long (permanent?) stalls in the block layer under heavy write > load. > ive experienced this problem with an intel nvme part that only has > intx and msix support. because we dont support msix yet we only > use intx on it. it appeared to lock up before this fix. > this has been tested on both that intel board and a samsung with msi. > this fix was based on work found in code by nonaka (dlg@) dev/pv ~ hyperv.c ~ hypervic.c ~ hypervicreg.h ~ hypervvar.h ~ if_hvn.c > Channel API users should pick the bufring size > hvn(4) needs to be able to put up to 128 RNDIS data messages on the > channel ring each pointing up to 16 fragments in addition to a small > number of RNDIS control messages. This simplifies accounting of > available TX descriptors in relation to the available space on the > channel ring and as a results prevents packet drops due to the ring > being full. (mikeb@) dev/usb ~ if_umb.c > Make kernel without INET6 compile again. > OK mikeb@ (bluhm@) net ~ if_vxlan.c > Make kernel without INET6 compile again. > OK mikeb@ (bluhm@) netinet ~ ip_carp.c > Accept CARP advertisement packets whose destination is not for multicast. > When both "carppeer" and "real mac-address" are used at once and the > BACKUP is to take over the new MASTER, the present MASTER receives > such packets. Found by and diff from nagasaka at iij. > ok henning bluhm (yasuoka@) netinet6 ~ raw_ip6.c > Fix bind(2)ing link local addresses to raw sockets by calling > in6_pcbaddrisavail() which does all the checking for us instead of > hand rolling half of it. > Input & OK bluhm@ > bluhm@ also points out that this relaxes the check for valid addresses > a bit, deprecated addresses become valid for raw sockets. This should > be fine, it brings raw sockets in line with udp/tcp sockets and it > might be particularly interesting for debugging purposes. (florian@) ~ raw_ip6.c > bluhm@ pointed out that addr->sin6_port should be forced to zero since > a port doesn't make sense for a raw socket. > I forgot to commit this hunk in the previous commit. Sorry about that. > (florian@) == usr.bin =========================================================== 04/05 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/usr.bin ssh ~ readconf.c > Fix logic in add_local_forward() that inverted a test when code was > refactored out into bind_permitted(). This broke ssh port forwarding > for non-priv ports as a non root user. > ok dtucker@ 'looks good' deraadt@ (jsg@) tmux ~ alerts.c > Do not handle activity or silence alerts in the current window, reported > by jasper@. (nicm@) == usr.sbin ========================================================== 05/05 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/usr.sbin makefs ~ makefs.c > convert usage() to just spit out the normal options list, > instead of trying to list all the fs-specific stuff; > help/ok natano (jmc@) tcpdump ~ print-ofp.c > Add missing OpenBSD keyword in the first line. (rzalamena@) =============================================================================== _______________________________________________ odc mailing list [email protected] http://www.squish.net/mailman/listinfo/odc
