Pull up following revision(s) (requested by hannken in ticket #2042):
sys/dev/fss.c: revision 1.116
Bracket I/O from or to the backing store with fstrans_start_lazy()
so the fss device doesn't block before the file system is suspended.
Should fix some deadlocks with revokation of ttys.
PR kern/59663 (ffs_snapshot_read -> uvm_fault (or pool page empty)
Pull up following revision(s) (requested by hannken in ticket #1320):
sys/dev/fss.c: revision 1.116
Bracket I/O from or to the backing store with fstrans_start_lazy()
so the fss device doesn't block before the file system is suspended.
Should fix some deadlocks with revokation of ttys.
PR kern/59663 (ffs_snapshot_read -> uvm_fault (or pool page empty)
Pull up following revision(s) (requested by hannken in ticket #409):
sys/dev/fss.c: revision 1.116
Bracket I/O from or to the backing store with fstrans_start_lazy()
so the fss device doesn't block before the file system is suspended.
Should fix some deadlocks with revokation of ttys.
PR kern/59663 (ffs_snapshot_read -> uvm_fault (or pool page empty)
Pull up following revision(s) (requested by rin in ticket #404):
crypto/external/bsd/openssh/dist/gss-genr.c: revision 1.17
crypto/external/bsd/openssh/dist/kexgen.c: revision 1.10
openssh: Fix up NetBSD RCSID. Otherwise, NFC
Pull up following revision(s) (requested by thorpej in ticket #403):
sys/dev/pci/isp_pci.c: revision 1.123
sys/dev/ic/isp_netbsd.h: revision 1.77
It has been observed in the wild that some individual instances of
ISP10[24]0 controllers don't seem to like getting a 64-bit DMA address
from our driver (at least, there seems to be some sort of incorrect
handling vis a vis the RQSTYPE_A64 message in isp_intr()).
Upon inspection, it was noted that in isp_pci_dmasetup(), the decision
to use RQSTYPE_T3RQS vs RQSTYPE_T2RQS was conditional on sizeof(bus_addr_t)
being larger than sizeof(uint32_t).
However, this isn't the correct predicate; instead, the choice should
be dependent on what the largest value that bus_addr_t will have. As
it happans, the PCI layer makes a distinction between 32-bit and 64-bit
DMA windows, and isp_pci's use of the regular "pa->pa_dmat" would ensure
that the DMA address always fits within 32-bits. Therefore, there was
[9 lines not shown]
Pull up following revision(s) (requested by thorpej in ticket #402):
sys/dev/ic/tulip.c: revision 1.215
Support for the 21142/21143 internal NWay block is structured like a
PHY driver, but it was overlooked when the ifmedia/mii layer gained a
locking protocol. Update it to avoid a recursive mutex acquisition
that can occur if the autonegotiation completes (very) quickly.
PR kern/60494
Pull up following revision(s) (requested by thorpej in ticket #401):
sys/net/if_media.c: revision 1.55
sys/net/if_media.h: revision 1.73
Expose ifmedia_match_locked(). This is a pre-requisite for fixing
the media locking error in the "tlp" driver described in...
PR kern/60494
Pull up following revision(s) (requested by skrll in ticket #400):
sys/arch/riscv/riscv/riscv_machdep.c: revision 1.51
sys/arch/evbarm/fdt/fdt_machdep.c: revision 1.113
The 2nd argument of fdt_memory_remove_range is a size. Fix the call sites
that get this wrong.
Prompted by the email to port-riscv from Alex Jokela
Pull up following revision(s) (requested by riastradh in ticket #398):
sys/dist/pf/net/pf.c: revision 1.88
pf(4): Avoid UAF of pf mbuf tag in fragment reassembly.
PR kern/60485: pf(4): uaf in fragment reassembly
Pull up following revision(s) (requested by riastradh in ticket #396):
sys/dev/hdaudio/hdaudio.c: revision 1.19
hdaudio(4): Apply access checks to ioctl routines.
This way the permissions on /dev/hdaudioN actually do something.
Sort includes while here.
PR kern/60492: hdaudio(4): setconfig is allowed unprivileged
Pull up following revision(s) (requested by riastradh in ticket #395):
libexec/ld.elf_so/tls.c: revision 1.32
libexec/ld.elf_so/tls.c: revision 1.33
libexec/ld.elf_so/tls.c: revision 1.34
ld.elf_so: Fix assertions for TLS variant I without _lwp_settcb.
In variant I, the TLS address is always
(uint8_t *)tcb + sizeof(struct tls_tcb) + obj->tlsoffset.
With _lwp_settcb, (uint8_t *)tcb + sizeof(struct tls_tcb) is aligned,
whereas without it, (uint8_t *)tcb is aligned. Consequently, with
_lwp_settcb, obj->tlsoffset must also be aligned, whereas without it,
obj->tlsoffset + sizeof(struct tls_tcb) must be aligned -- and hence,
without _lwp_settcb, obj->tlsoffset must be congruent to
-sizeof(struct tls_tcb) modulo the alignment.
[29 lines not shown]
Pull up following revision(s) (requested by joe in ticket #394):
sys/dev/raidframe/rf_disks.c: revision 1.96
sys/dev/raidframe/rf_disks.c: revision 1.97
PR kern/60477
Use the right assertions to ensure that all IOs are suspended.
zero indicates that IOs resume.
PR kern/60477
Use greater than zero instead of strictly one
PR install/59195 and PR port-evbmips/60302: fix hashes in gzimg directories
Hashes in gzimg directories for some ports either correspond to the
wrong (i.e., uncompressed) files or do not cover all files. These
problems are described in PR install/59195 and PR port-evbmips/60302,
respectively. The potential solution offered for PR 59195 does not
work if multiple image files exist in a single gzimg directory.
The correct solution involves three aspects: (i) make checksums
independently of mkimage, which allows a common list of hashes in
etc/Makefile, (ii) for directories producing more than one compressed
image, make checksums in individual temporary files, one per
compressed image, to combine at the end, and (iii) add make targets to
allow parallelism across multiple images within each directory.
This patch implements those steps for the evbarm, evbmips, evbppc, and
riscv ports, which were the only ones previously using mkimage for
checksums.
Add a MKRAID build option, which defaults to yes everywhere except
m68000, which controls building of RAID-related utilities (including
hardware RAID controllers).
build.sh: consider tools/make/configure and tools/make/*.in
When deciding if nbmake is out-of-date the files in tools/make
should be considered too.
Reviewed by: riastradh