Reject empty CNAMEs in gethostbyname(3) / getaddrinfo(3).
An empty string is arguably not a correct hostname (even though
res_hnok accepts it). More worrisome though is software not expecting
this and making mistakes. In practice this cannot happen unless the
resolver lies to us.
OK deraadt, jca
Rework non-functional CA statistics accounting
This also fixes a subtle defect: non-functional CAs were counted
towards the repository those broken CAs were pointing towards,
instead of being counted towards the repository that contained the
broken CA.
Attributing the non-functional CA to the issuing parent makes more
sense, because the issuing parent can actually do something about it,
for example, by revoking the non-functional child CA.
OK tb@
The busdma API requires use of typedefs and pointers, and generally
hides the structures and their fields. Obviously a few piece of MD
code need to setup these structures, and therefore the structure names
became MD with prefixes like arm64_ or sparc_ or machine_. amd64 did
not follow this pattern of using a prefix. The hibernate code for
side-effect-free driver setup needs to handle the real structures.
This is MI code and is being done carefully without becoming an API
leakage concern.
Other ways of fixing this are worse, so we decide these MD prefixes
are a bad idea. This removes the prefix for arm64 so that hibernate
work can happen.
ok kettenis mlarkin mglocker
O_CLOFORK was added here by accident.
I am becoming convinced quickly that the invention of O_CLOEXEC
and later acceptance into standard by austin group, is a serious
mistake and we should rip the support out of the kernel.
ok jca
don't deactivate the whole usb device if attach can't find endpoints.
r1.128 let umsm be greedy and attach to more interfaces on the same
usb device, rather than just the first 4 interfaces. Qeuctel EM060K-GL
modems have interfaces at 8 and 12, the first of which is umb but
the 12th used to attach as ugen. umsm now tries to attach to interface
12, but because it doesn't have the right set of endpoints the
attach hits an error path and disable the device.
this removes the usbd_deactivate call in the that error path, which
leaves all the other endpoints working.
this follows the advice of quectel. kevlo and i are talking about
moving to the linux model of maintaining a much more comprehenive
table of modems and which endpoints are supported, but this is an
improvement in the mean time.
ok kevlo@ deraadt@
drm/amd/display: Use krealloc_array() in dal_vector_reserve()
From Harry Wentland
e09689286385a66311ac6922af95339d7a3cef8d in linux-6.18.y/6.18.36
da48bc4461b8a5ebfb9264c9b191a701d8e99009 in mainline linux
drm/amd/display: Fix out-of-bounds read in dp_get_eq_aux_rd_interval()
From Harry Wentland
454d3b3d499c18373f8960d31aea48338a3ca9e0 in linux-6.18.y/6.18.36
e8b4d37eba05141ee01794fc6b7f2da808cee83b in mainline linux
drm/amd/display: Fix NULL deref and buffer over-read in SDP debugfs
From Harry Wentland
bb6f705b73b5f191f14ad004e2c8c4b615806187 in linux-6.18.y/6.18.36
adf67034b1f61f7119295208085bfd43f85f56af in mainline linux
drm/amd/display: add missing CSC entries for BT.2020 for DCE IPs
From Leorize
c000da79df787097abdfb1cb4bf94af8151a566a in linux-6.18.y/6.18.36
6590fe323ce2807f5d9454e7fccf3fab875d4352 in mainline linux
drm/amd/display: Clamp VBIOS HDMI retimer register count to array size
From Harry Wentland
3f32d52ec604c659725d865cf8cc6a17a33f9c6a in linux-6.18.y/6.18.36
fb0707ce00eef4e2d60c3020e1c0432739703e4a in mainline linux
drm/amd/display: Clamp HDMI HDCP2 rx_id_list read to buffer size
From Harry Wentland
1906064d50d194a145486e5caf3db3e708b6f6ef in linux-6.18.y/6.18.36
f0f3981c43b32cadfe373d636d9e9ca522bb3702 in mainline linux
drm/amd/display: Bound VBIOS record-chain walk loops
From Harry Wentland
0e56f460bddb397fa9a8e6faf7ae7eaa86953eb1 in linux-6.18.y/6.18.36
ff287df16a1a58aca78b08d1f3ee09fc44da0351 in mainline linux
drm/amd/pm: smu_v14_0_0: use SoftMin for gfxclk in set_soft_freq_limited_range
From Priya Hosur
57607fe55e6d598bc58c15d1143ba7a266b17103 in linux-6.18.y/6.18.36
03b70e0d8aa26bab89a0f1394c1c80a871925e42 in mainline linux
drm/amd/pm: mark metrics.energy_accumulator is invalid for smu 14.0.2
From Yang Wang
932642791cb1513118847370f4cbe09d69400eed in linux-6.18.y/6.18.36
ee193c5bbd5e2b56bbeb54ef554414b43a6fc896 in mainline linux
drm/amd/pm: fix smu13 power limit default/cap calculation
From Yang Wang
8979ded4d8997cdaba94a6c58be2ad069c720ee8 in linux-6.18.y/6.18.36
bb204f19e4a115f094a6a3c4d82fcf48862d0766 in mainline linux
drm/amdgpu: set noretry=1 as default for GFX 10.1.x (Navi10/12/14)
From Vitaly Prosyak
39b5397bf8de2c5f110a0599707a7b1e0dbc6bf3 in linux-6.18.y/6.18.36
e47b0056a08dc70430ffc44bbf62197e7d1ff8ea in mainline linux
drm/amdgpu: restart the CS if some parts of the VM are still invalidated
From Christian Koenig
fcd51a085e9a0d01315606e9d2dad1732563faf2 in linux-6.18.y/6.18.36
40396ffdf6120e2380706c59e1a84d7e765a37b6 in mainline linux
drm/amdgpu: fix waiting for all submissions for userptrs
From Christian Koenig
68455b117258243e73b9697d15a570158ad93e1c in linux-6.18.y/6.18.36
58bafc666c484b21839a2d27e923ae1b2727a1df in mainline linux
drm/amdkfd: Fix buffer overflow in SDMA queue checkpoint/restore on GFX11
From Andrew Martin
d3efcadfe3eea5b4263b8f2d4463b15c9fc46a64 in linux-6.18.y/6.18.36
352ea59028ea48a6fff77f19ae28f98f71946a80 in mainline linux
drm/amdkfd: fix NULL dereference in get_queue_ids()
From Muhammad Bilal
72e259a32084c42816152c346096d2edd4213e23 in linux-6.18.y/6.18.36
2bd550b547deabef98bd3b017ff743b7c34d3a6d in mainline linux
drm/gem: Try to fix change_handle ioctl, attempt 4
From Simona Vetter
c0639ede2f24ac224b2079cd35ecd5fd8ad4e3cd in linux-6.18.y/6.18.36
1a4f03d22fb655e5f192244fb2c87d8066fcfca2 in mainline linux
drm/amd/display: Reject gpio_bitshift >= 32 in bios_parser_get_gpio_pin_info()
From Harry Wentland
edabfe80e34efc468abb3ae5586d71830a1ae585 in linux-6.18.y/6.18.36
49c3da65961fe9857c831d47fa1989084e87514a in mainline linux
drm/i915/gem: Fix phys BO pread/pwrite with offset
From Joonas Lahtinen
32d4c5d328a3ff995420f4f85163e1e403f43628 in linux-6.18.y/6.18.36
d21ad938398bca695a511307de38a65889e3b354 in mainline linux
drm/i915/edp: Check supported link rates DPCD read
From Nikita Zhandarovich
dda720b2928d162ad76273cd35d881bc69b71600 in linux-6.18.y/6.18.36
2673cefa99ca918e7ac5b0388ff578a83656c896 in mainline linux
raise the size of amd64 kernel virtual address space from 4G to 512G
This allows mapping all of VRAM on recent Radeon cards where the
PCI BAR size is the full (> 4G) size.
Reported by Petre Rodan, who has a Radeon RX 6800 with 16G of VRAM.
feedback from claudio@ ok kettenis@
change iommu_map_sgtable() return type from size_t to ssize_t
Should be signed as error paths return -ENOMEM. Linux made the same
change a few years ago.
Do `ip_shuffle' initialization at boot time. ip_randomid() called in the
hot path so we want to keep it lockless, but avoid access to uninitialized
data.
ok deraadt