| Age | Commit message (Collapse) | Author |
|---|
|
To make this intent more obvious, group the call to the TLB flush function
togther with the call that inserts/updates/removes a PTE.
Remove an incorrect comment related to this.
ok patrick@
|
|
The SMC implements a lot of functionality. For now the driver only
implements a bunch of sensors. This is a small subset of the sensors
that are made available by the SMC as we don't know what measurments
are provided for most sensors.
ok patrick@
|
|
|
|
the status ring entries are 8 bytes/64bit, and depending on the
type of entry it has fields all over the place. this loads the
descriptor with a single 64bit read, and then shifts and masks the
bits out of it depending on the type of descriptor. this looks
cleaner for the tx completions in particular.
|
|
CID 1510483
ok kevlo@
|
|
OK deraadt@ phessler@
|
|
OK deraadt@ phessler@
|
|
unveil(2). It is not set for nodes that are added as a result of a file
being added via unveil(2). Use this flag to test if backtracking should
be done or not. Also introduce UNVEIL_MASK which checks if any user flags
are set and is used to properly return EACCES vs ENOENT.
This fixes a problem where unveil("/", "r") & unveil("/usr/bin/id", "rx")
cause an error when read accessing "/usr/bin". It also makes sure that
unveil(path, "") will return ENOENT for any access of anything under path.
Reported by and OK semarie@
|
|
feedback and ok tb@ jmc@ ok ratchov@
|
|
Delete comment referring to it
ok jsg@
|
|
|
|
|
|
BOOT kernels do not print kernel messages, and currently there is no
way to change this at runtime. Remove the verbose device information
to save some space.
|
|
|
|
Rework the tx path to use the consumer and producer positions to work out
the number of slots available, and to put packets on the ring until fewer
than DWXE_NTXSEGS slots are left, making dwxe_start() and dwxe_txeof()
work independently. While here, only write to DWXE_TX_CTL1 once
per call to dwxe_start() rather than once per packet.
Adjust the rx interrupt path to check the number of slots in use and
return slots once per interrupt.
Add interrupt and ifq barriers before taking the interface down.
With all of this done, we can mark dwxe(4) mpsafe.
tested on arm64 (a64 sopine) by mlarkin@ and armv7 (h2+) by me
ok dlg@
|
|
U-Boot to initialize the hardware for us, but it is better if we can cope
with this ourselves.
ok patrick@
|
|
interface. If this operation fails (probably due to missing firmware),
we must undo changes to the SIOCSIFXFLAGS xflags.
ok stsp.
|
|
thumb code calling arm code, no need to put these input sections at the text
output section.
ok jsg@ kettenis@
|
|
|
|
|
|
pass in the already read hibernate_info instead of reading it again.
ok deraadt@
|
|
it's the 'b' slice and (sanity) check against the partition count.
Also, make the "is union hibernate_info too large?" a compile time
check.
ok deraadt@
|
|
macros to make kernel build again, same diff as armv7.
ok kettenis visa
|
|
macros to make armv7 build again.
ok deraadt@
|
|
|
|
uses the same green sardine firmware as cezanne
|
|
From Alex Deucher
b8553330a07749e488d143b5704adf1042fd7c0a in linux 5.10.y/5.10.90
5e713c6afa34c0fd6f113bf7bb1c2847172d7b20 in mainline linux
|
|
enabled
From chen gong
28863ffe21ff711d5109e3c208676258bdec3a1f in linux 5.10.y/5.10.90
b7865173cf6ae59942e2c69326a06e1c1df5ecf6 in mainline linux
|
|
|
|
the bridge when present on FDT platforms. Needed on platforms like the
Apple M1 to make sure the PCI bus numbers match the IOMMU setup required
by the device tree.
ok patrick@
|
|
boundaries: hppa has 8-byte PLT entries that sometimes do that.
ok kettenis@
|
|
pointed out by + ok millert@
|
|
ok jsg kettenis
|
|
This bug caused the driver to read block ack request information sent
by firmware from the wrong offset. The driver flushes buffered frames
and moves its Rx block ack window based on this information. Possible
consequences of this bug are packet loss or even stalled traffic if
the Rx BA window gets out of sync between driver and firmware. Though
this effect might get cancelled out when the driver re-syncs the BA
window in its regular Rx code path.
Spotted by Christian Ehrhardt.
|
|
ok deraadt@
|
|
everyone else seems to use ETHERTYPE_EAPOL, and as a bonus it also
appears to be more correct.
ok deraadt@ stsp@
|
|
this replaces the use of a register in txeof to figure out how much
of the ring has been completed by the hardware. that register isn't
reset when an interface is taken down and brought up again, which
messes with the calculations of free slots on the ring, which in
turn messes up the ability to transmit packets.
this means the watchdog can't know where the hardware is up to
anymore, so just restartthe chip if the watchdog fires.
it seems to be important to start the tx ring in msk_init on the
0th ring entry that we use to reset the high address value to 0.
we still fill the 0th descriptor, but we let the first msk_start
call post it for us when a packet goes on the ring.
a slight tweak and ok jmatthew@
|
|
OK kettenis@
|
|
|
|
to includes pcivar.h because it inspects vid/pid tables
|
|
this file
|
|
resume. This fixes setups where a umass device no longer attaching
at resume results in a softraid device being renumbered so the
hibernate-time device is no longer correct
ok mlarkin@ jsing@
|
|
This fixes issues on the M1 Macs where the PCI probe done by Xorg
breaks the WiFi chip.
ok patrick@
|
|
The minimum length and the maximum length required were both too low,
due to an error in accounting for the 4-byte packet length+flags header.
Patch by Christian Ehrhardt
|
|
I ported a NULL pointer check from iwlwifi rxmq.c which cannot
happen in our version of this code. Instead we need to check
the BA session ID to determine whether a BA session is valid.
Patch by Christian Ehrhardt.
|
|
correct property. While there adjust acpi_getpropint() as well to
increase similarity with acpi_getprop().
ok kettenis@
|
|
|
|
ok visa@
|
|
|
|
trees. ipsp_ids_lookup() returns `ids' with bumped reference
counter. original diff from mvs
ok mvs
|
