Re: rtsock refactoring

On Sat, Jan 21, 2017 at 07:31:20AM +0100, Claudio Jeker wrote: > On Sat, Jan 21, 2017 at 01:28:02AM +0100, Claudio Jeker wrote: > > On Fri, Jan 20, 2017 at 02:51:52AM +0100, Claudio Jeker wrote: > > > I sent this diff out some time ago and would really like to get this in. > > > This is one step

Re: ipsec_common_input() & splsoftnet()

ok Martin Pieuchot(m...@openbsd.org) on 2017.01.23 10:44:59 +1000: > This function is only called by udp_input() so at IPL_SOFTNET. Assert > that and get rid of a recursive splsoftnet()/splx() dance. > > ok? > > Index: netinet/ipsec_input.c >

less

This change causes less(1) to open /dev/tty slightly earlier, or fallback to stderr. As a result the TIOCGWINSZ operation can be done against the most likely tty descriptor. As a side effect, that tty descriptor will be opened before the first pledge call. Anyone see a downside? Index:

arm64: distrib for ramdisk kernel

Hi, this diff provides the infrastructure to compile a ramdisk kernel for arm64. It's been copied from armv7 and stripped from u-boots, device trees and mkuboot. ok? Patrick diff --git a/distrib/arm64/Makefile b/distrib/arm64/Makefile new file mode 100644 index 000..bb5d606b329 ---

Special casing fd's opened before pledge(2)

There has been developer pressure to permit an increasing number of ioctl's to pledged programs. The problem is that providing a specific ioctl under a promise to one program, means it becomes supplied to all other programs that make that promise. There is no discrete method to differentiate

arm64: distrib/special installboot

Hi, this makes installboot compile for arm64 at the RAMDISK build. ok? Patrick diff --git a/distrib/special/installboot/Makefile b/distrib/special/installboot/Makefile index 760daf0e44e..bfff63b11e4 100644 --- a/distrib/special/installboot/Makefile +++ b/distrib/special/installboot/Makefile

Re: use ifqueue packet statistics

On 22/01/17(Sun) 20:38, David Gwynne wrote: > ifqueues count packets, therefore if_enqueue doesnt have to. > > this adds reading of the ifq counters into the if_data values so > userland still looks right. > > this is another step toward multiple tx queues. I like it, some comments below. >

Re: use ifqueue packet statistics

On Mon, Jan 23, 2017 at 10:10:23AM +1000, Martin Pieuchot wrote: > On 22/01/17(Sun) 20:38, David Gwynne wrote: > > ifqueues count packets, therefore if_enqueue doesnt have to. > > > > this adds reading of the ifq counters into the if_data values so > > userland still looks right. > > > > this is

add support for multiple transmit queues on interfaces

the short explanation is that this lets interfaces allocate multiple ifq structures that can be mapped to their transmit rings. the mechanism for this is a driver calling if_attach_queues() after theyve called if_attach(). the long version is that this has if_enqueue access an array of ifqueues

ipsec_common_input() & splsoftnet()

This function is only called by udp_input() so at IPL_SOFTNET. Assert that and get rid of a recursive splsoftnet()/splx() dance. ok? Index: netinet/ipsec_input.c === RCS file: /cvs/src/sys/netinet/ipsec_input.c,v retrieving

Re: split pledge "ioctl" into "bpf" and "tape"

> From: Theo de Raadt > Date: Sun, 22 Jan 2017 20:52:14 -0700 > > Early during pledge development the "ioctl" promise was a kitchen > sink of options until we could differentiate use cases, identify > common patterns, and then create domain-specific promises. > > only 4

Re: rtwn: fix iq calibration

Stefan Sperling writes: > FreeBSD committed a couple of interesting things in r307529 > https://svnweb.freebsd.org/base?view=revision=r307529 > Unfortunately the commit is a giant patch bomb which makes > it hard to pick out individual fixes :-( > > This patch extracts what I

uvm: guard unused function for RAMDISK kernel

Hi, on a RAMDISK kernel this function is not used, thus clang complains. Simply guard it with SMALL_KERNEL, like the rest of that part of the code. ok? Patrick diff --git a/sys/uvm/uvm_addr.c b/sys/uvm/uvm_addr.c index af33700273b..4527e1fda78 100644 --- a/sys/uvm/uvm_addr.c +++

Re: uvm: guard unused function for RAMDISK kernel

On Mon, Jan 23, 2017 at 10:20:39AM +1000, Patrick Wildt wrote: > Hi, > > on a RAMDISK kernel this function is not used, thus clang complains. > Simply guard it with SMALL_KERNEL, like the rest of that part of the > code. > > ok? > > Patrick > As requested by deraadt@, simply move the function

split pledge "ioctl" into "bpf" and "tape"

Early during pledge development the "ioctl" promise was a kitchen sink of options until we could differentiate use cases, identify common patterns, and then create domain-specific promises. only 4 cases remain of "ioctl" remain: - pax/tar/cpio experience great difficulty finding tape drives

rtwn: fix iq calibration

FreeBSD committed a couple of interesting things in r307529 https://svnweb.freebsd.org/base?view=revision=r307529 Unfortunately the commit is a giant patch bomb which makes it hard to pick out individual fixes :-( This patch extracts what I believe "fix IQ calibration bug" refers to. The current

define pvbus in conf/files

Hi, when fixing my virtio config breakage, I found the following comment in sys/conf/files, followed by a few bus defines: ---snip--- # Attributes which machine-independent bus support can be attached to. # These should be defined here, because some of these busses can have # devices which

Re: usbdevs: logitech: cleanup and new ids

On 1/23/17, Jonathan Gray wrote: > On Mon, Jan 23, 2017 at 01:56:20AM -0500, James Hastings wrote: >> Hello @tech >> >> Recently found a pile of old Logitech webcams to test with uvideo(4). >> s/QUICKCAM/QKCAM. >> shorten device strings. >> Add many new Logitech device ids. >> >>

Re: usbdevs: logitech: cleanup and new ids

On Mon, Jan 23, 2017 at 01:56:20AM -0500, James Hastings wrote: > Hello @tech > > Recently found a pile of old Logitech webcams to test with uvideo(4). > s/QUICKCAM/QKCAM. > shorten device strings. > Add many new Logitech device ids. > > dev/usb/usbdevs > dev/usb/uaudio.c > dev/usb/uvideo.c USB

Re: add support for multiple transmit queues on interfaces

hrvoje popovski hit a problem where the kernel would panic under load. i mistakenly called an interfaces qstart routine directly from if_enqueue rather than via the ifq serializer. this meant that txeof routines on network cards calling ifq_restart would cause the start routine to run

usbdevs: logitech: cleanup and new ids

Hello @tech Recently found a pile of old Logitech webcams to test with uvideo(4). s/QUICKCAM/QKCAM. shorten device strings. Add many new Logitech device ids. dev/usb/usbdevs dev/usb/uaudio.c dev/usb/uvideo.c Index: dev/usb/uaudio.c

use ifqueue packet statistics

ifqueues count packets, therefore if_enqueue doesnt have to. this adds reading of the ifq counters into the if_data values so userland still looks right. this is another step toward multiple tx queues. ok? Index: if.c === RCS