update the match code for quectel devices.
this is based on information in the Quectel UMTS LTE 5G Linux USB
Driver User Guide V3.3 documentation, which is the latest version
i can find.
it works on EC200A devices, which benefit from this change, and on
EC25s, which were working before this change and still work after.
implement IFCAP_TSO support.
this allows the stack to pass large tcp frames between rdomains.
note that rport doesn't advertise LRO support, which mostly means
that LRO is not something that can be disabled on rport interfaces.
as an l3 tunnel interface, there's no topology that an rport interface
can be set up in where disabling LRO is needed, unlike l2 Ethernet
interfaces. coordinating disabling LRO on an rport and then clearing
the TSO flags on the partner rport interface is not worth the effort
to actually implement the capability isn't worth it.
Collect sysctl kern.maxpartitions into MAXPARTITIONS variable early on
Then use this in MAKEDEV.common and MAKEDEV.mi for various places to
create the right number of partitions (a-p or a-zA-Z) as required.
reviewed mostly by krw and sthen
There could still be some glitches. This is being done step by step,
because I hope to cross some architectures at different times because
some of them have "emulated disklabels" which could be tricky.
expose some service information as kstats
this makes the uplink and downlink speeds visible as kstats. this
is in addition to the signal strength metrics that are already
exposed as kstats.
reduce default so-sndbuf in unbound to avoid a warning.
in 1.24.0, upstream started to override the OS default and warn at
startup if not possible, but they picked a value (4MB) which is above
the 2MB maximum that we permit.
found by landry@, ok landry bluhm
from bluhm:
"Our UDP stack has no UDP socket buffer for sending. Each packet
is directly passed to ip_output(). So basically this value is a
per packet size limit."
Rename ManifestRef to ManifestInstance
This is preparation step for an upcoming change to the CCR format in
which the structure containing manifest metadata will gain an extra
field (and thus not align with the ManifestRef structure from
draft-spaghetti-sidrops-rpki-erik-protocol)
OK tb@
drm/amdgpu: Enable MES lr_compute_wa by default
From Mario Limonciello
5980a35c9d138804251e50788c1e8137028a47ac in linux-6.12.y/6.12.52
1fb710793ce2619223adffaf981b1ff13cd48f17 in mainline linux
drm/amd/include : Update MES v12 API for fence update
From Shaoyun Liu
2dedc6b77bf82dae6cf6a9d4d5ac44526be9a605 in linux-6.12.y/6.12.52
15d8c92f107c17c2e585cb4888c67873538f9722 in mainline linux
drm/amd/include : MES v11 and v12 API header update
From Shaoyun Liu
ff2b82286fc8e6fc6caf9072009d78f7c05db32e in linux-6.12.y/6.12.52
1c687c0da9efb7c627793483a8927554764e7a55 in mainline linux
drm/amd : Update MES API header file for v11 & v12
From Shaoyun Liu
bfd0bec4cb2c0b9e30decc84ffc269ba91ee6906 in linux-6.12.y/6.12.52
ce4971388c79d36b3f50f607c3278dbfae6c789b in mainline linux
Revert rev 1.109:
| ixl/ice(4): use 128 segments for DMA maps of TSO packets
|
| This avoids unnecessary m_defrag() calls and gain some performance.
|
| with tweaks for kettenis and bluhm
|
| ok kettenis, bluhm
This seems to cause oactive stalls and ixl is too common to have it broken
like this. I prefer unnecessary m_defrag() calls over hanging interfaces.
use strtonum() instead of atoi(), and error out for bad numbers
This generates a host-order number, so the ntohs() for getservbyport()
was wrong, that should always have been htons(). The transform is
the same, but misleading.
ok tb