uncm(4) update from Jan Schreiber and Mikolaj Kucharski
- remove unneeded splnet() calls
- during Tx, drop packets which are too big for a single NTB
- validate NCM_FORMAT before using this device-supplied value
- use the ifqueue(9) API to enqueue packets
- be more spec-compliant while checking wNdpOutAlignment
- close the control pipe before calling usb_rem_wait_task(9), not after
Correct the description: the argument is effectively boolean,
and the only values that can be set are 0 and 3.
Minor polishing while here: use .Xr the first time "sysctl" occurs,
use .Rv below RETURN VALUES, add a few .Xrs below SEE ALSO,
and use the standard section name "CAVEATS" rather than "WARNING".
OK miod@; triggered by a question from Alejandro Colomar <alx at kernel.org>
rpki-client: close directory fds in process_start()
After changing the working directory to the cache directory, none of the
children needs access to either of the two directory fds in the parent.
Also close the cachefd when it is no longer needed in the main process to
match what was done to the output directory. Apart from some consistency,
the benefit of doing this is less obvious to me.
discussed with claudio and deraadt
ok claudio
ocspcheck: limit HTTP response body size
Reject OCSP response bodies larger than 20 KB before growing the body
buffer. A responder could otherwise stream data indefinitely and cause
ocspcheck to exhaust memory.
From Frank Denis, also reported by Aziz ELBELAYCHY
ok tb@
ocspcheck: limit HTTP response header size
Reject response headers that would exceed 64 KB before growing the
header buffer. A responder could otherwise stream data without a
header terminator and cause ocspcheck to allocate memory indefinitely.
From Frank Denis, also reported by Aziz ELBELAYCHY
ok tb@
ocspcheck: remove disabled HTTP test program
Remove the unused main() function at the end of http.c. It has been
excluded from the build with #if 0.
Pointed out by tb@
ok tb@
ocspcheck: select TLS based on the OCSP URL scheme
Preserve the URL scheme when parsing the OCSP responder address and pass
it to the HTTP client. Previously, the client selected TLS only when
the port was 443, so an https:// URL with a different port sent the
OCSP request in cleartext.
Use the scheme for the verbose connection message as well.
Reported by Aziz ELBELAYCHY
ok tb@
Revert rev. 1.134 of uvm_vnode.c because it introduces a reference race
uo_refs is not really serialized by the vmobjlock and at least in
uvm_vnp_sync() it depends on the KERNEL_LOCK for that.
The optimisation done in 1.134 predates the rwlock rework done by dlg@
and is probably no longer important. The vmobjlock is still contended
but the wakeup storm resulting in massive KERNEL_LOCK spin is gone.
Reported by Acts1631
OK kettenis@ deraadt@
drm/amd/display: Don't replace a sink mode that shares the native totals
From Adrian Betschart
f94a12ddd3184280c4a1d0ce4c8be7d36d0a52a3 in linux-6.18.y/6.18.57
54c998c55fc9cbc391b350d67b86e0c2a0748f1a in mainline linux
drm/dp: don't mark the AUX backlight enabled when enabling failed
From Oleg Keri
5b26dbb27c0cb0e1b92fc0626c6ae052da8ff7d6 in linux-6.18.y/6.18.57
707ce102521496c67860d81221b20c56d8255f2f in mainline linux
drm/amdgpu: skip the VMID 0 flush for VRAM
From Arunpravin Paneer Selvam
30dacb56f07283a1acb353ce81baef31f85396b0 in linux-6.18.y/6.18.57
87ceb8cba73d0b3c4025ff42495bccd8164acaed in mainline linux
drm/amd/display: Fix dangling pointer in plane reset function
From Evgenii Burenchev
3d8391d3149a1aeb93505a4f9fb9ed8364882c42 in linux-6.18.y/6.18.57
98cad4bd1443975d972f4c7f705980da03722a22 in mainline linux
drm/amd/display: Fix dangling pointer in CRTC reset function
From Evgenii Burenchev
3a03827bb3cda1527aa923f18160c449f2b4cf2c in linux-6.18.y/6.18.57
0aeed866cb938943908c3ba46422128e49d2d080 in mainline linux
drm/amd/display: Resize MST HDCP per-connector arrays to 32
From Harry Wentland
562d24ac214ba9530d88841088eb6adce5c537b5 in linux-6.18.y/6.18.57
261e0fe4e2c99f687114b64b10b98db964b475f4 in mainline linux
drm/amd/display: disable replay when crc source is enabled
From Ray Wu
b5896fc269e9f3b3b4a67a7ce78872bc3bbd3b25 in linux-6.18.y/6.18.57
a036db3953140d5c2816b056aa3e73b8e98d9bc0 in mainline linux
drm/amdgpu: force complete the KIQ ring fences on reset
From Jesse Zhang
c26cbd5c33cd72daa03079235de60688a3c97c52 in linux-6.18.y/6.18.57
fd65d1742992361fc2201ecb4e43411e6e417fcb in mainline linux
drm/amdgpu: force complete the MES ring fences on reset
From Jesse Zhang
7d722821e2f626c4e34c8510f0ce1d6f10c77a3b in linux-6.18.y/6.18.57
48dc279c3010ac8f91b1845b2abb3a1e9943a0f5 in mainline linux
drm/amdkfd: Add bounds check for CRAT subtype length
From William Palacek
69164d6bdad5d6657e51436a3f602822f620a46a in linux-6.18.y/6.18.57
6e7566ba4739dd573c331adde1c96690f7a567bd in mainline linux
drm/amdkfd: Fix infinite loop parsing CRAT with zero subtype length
From Yongqiang Sun
39841db9322cddf7bd1dbcacdffc791475c7b02f in linux-6.18.y/6.18.57
961323c26ad4c895e3b0ea1711fc41dfd6368c12 in mainline linux
drm/amd/display: check if dml21_add_phantom_plane() is successful
From Alex Deucher
bea1f73ad69eec36b757adb8e63beb3092f862ee in linux-6.18.y/6.18.57
000acb4ce7fb9feba3072ce468ad681f6585cd5d in mainline linux
drm/amd/display: Fix divide-by-zero in calculate_mcache_setting on zero viewport
From George Zhang
23a541c31b45fa813b33269376f6bd53535429a8 in linux-6.18.y/6.18.57
f327e389c07cfc3a2f6ff54f6214e1a52d457edc in mainline linux
