On Sun, 5 Aug 2018 00:52:32 -0700, Ori Bernstein wrote:
> This change introduces a 'struct virtio_backing' which makes the
> disk i/o pluggable, providing 'backing->{pread,pwrite}' calls that
> can be replaced by different disk i/o drivers.
>
> This is necessary preparation for adding qcow2 supp
As per X.690, "the end-of-contents octets shall be present if the length is
encoded as specified in 8.1.3.6, otherwise they shall not be present", i.e.
only used with indefinite length encoding. Since we do not support indefinite
length encoding, I thought it may make sense to remove some correspon
Somewhere along the line of file descriptor code refactoring the
refcounting for dmabuf file descriptors got botched. The result is
that the files never die and the associated graphics objects never get
freed. Eventually you'll run out of memory.
I noticed this after switching to a Mesa 17.3.9 b
Some access points have a feature called "band steering" where they
will try to push clients from 2 GHz channels to 5 GHz channels.
Such APs keep track of probe-requests from the same client on
different bands. If a client sends probe-requests on both 2 GHz
and 5GHz channels, and then attempts to a
iwm switches the current BSS channel for each input frame.
The reason for this is that with most of our drivers the
net80211 stack zaps through all channels one-by-one during
scans, and hence expects the BSS channel to match the
currently scanned channel.
However, iwm offloads scans to the firmwa