Re: mfi(4): Make "bioctl -R" work after hot swapping

Hi Jonathan Matthew, From: YASUOKA Masahiko Subject: Re: mfi(4): Make "bioctl -R" work after hot swapping Date: Tue, 24 Apr 2018 20:20:40 +0900 (JST) > On Thu, 29 Jun 2017 17:14:41 +0900 (JST) > FUKAUMI Naoki wrote: >> Currently "bioctl -R" works only if disk state is "Offline" (set by >> "bioc

Re: Change CMakeLists.txt in LibreSSL to use target_include_directores

Question about the PUBLIC status of the ../include/compat headers in CMakeLists.txt. I wrote the target_include_directories calls to include ../include/compat in each of the targets and marked them PUBLIC, but I’m wondering if that will cause conflicts with system headers like time.h and if the

Re: Fewer sofree()

On Mon, Jun 04, 2018 at 03:25:17PM +0200, Martin Pieuchot wrote: > For pfkey and routing sockets, calling sofree() in pr_detach is a noop. > The reason is that `SS_NOFDREF' hasn't been set at this point so the > socket won't be freed. > > So I'd like to remove the sofree() and add an assert instea

Re: vfs_cache.9: sync with

On Fri, Jun 01 2018, Klemens Nanni wrote: > On Sun, May 27, 2018 at 04:37:52PM +0200, Klemens Nanni wrote: >> Spotted and reported on IRC by Georg Bege , >> vfs_cache(9) lacks way behind beck's "Namecache revamp" from 2009. >> >> This diff syncs the manual with sys/sys/namei.h and sys/kern/vfs_ca

Re: route: regress: Allow specifying binary via ROUTE

On Sun, Jun 03 2018, Alexander Bluhm wrote: > On Sun, Jun 03, 2018 at 12:26:18AM +0200, Klemens Nanni wrote: >> With `make ROUTE=/usr/obj/sbin/route/route' I can test my patched >> version without having to modify PATH or the regress Makefile. >> >> pfctl already does it that way. > > This is th

Re: netstat: zap unused maxmif

On Sun, Jun 03 2018, Klemens Nanni wrote: > Unused since introduction in 1.17 from 2015. > > OK? ok jca@ > Index: mroute6.c > === > RCS file: /cvs/src/usr.bin/netstat/mroute6.c,v > retrieving revision 1.23 > diff -u -p -r1.23 mroute

Re: route: zap unused sockaddr

On Sun, Jun 03 2018, Alexander Bluhm wrote: > On Sun, Jun 03, 2018 at 01:05:32AM +0200, Klemens Nanni wrote: >> No object change. >> >> OK? > > usr.bin/netstat/show.c has the same code. They should stay in sync. > Please change it there, too. Then OK bluhm@. Agreed, please amend netstat too (o

Re: [patch] Add kvm_close in mib_hrsystemprocs function

On Mon, Jun 04 2018, Gerhard Roth wrote: > On Thu, 31 May 2018 17:40:36 +0800 Nan Xiao wrote: >> Hi Gerhard, >> >> Thanks for your reply! >> >> Yes, if no "kvm_close(kd);", there will be resource (memory, file >> descriptor) leak. So hope you can commit it, thanks! >> >> >> On 5/30/2018 4:49

Re: drop unused BUMPTIME macro

> Date: Mon, 4 Jun 2018 12:20:29 -0500 > From: Scott Cheloha > > miod@ dropped the last usage of BUMPTIME circa 5.3. > > ok? ok kettenis@ > Index: sys/kern/kern_clock.c > === > RCS file: /cvs/src/sys/kern/kern_clock.c,v > retrievi

Re: Towards per-socket locks: sofree() & sounlock()

On Mon, Jun 04, 2018 at 03:56:02PM +0200, Martin Pieuchot wrote: > Diff below adds a 'struct socket *' argument to sounlock() in order to > prepare the stack for per-socket locks. > > That means sofree() will now unlock a given socket before freeing it. > But since we do not want to not release th

drop unused BUMPTIME macro

miod@ dropped the last usage of BUMPTIME circa 5.3. ok? -- Scott Cheloha Index: sys/kern/kern_clock.c === RCS file: /cvs/src/sys/kern/kern_clock.c,v retrieving revision 1.94 diff -u -p -r1.94 kern_clock.c --- sys/kern/kern_clock.c

vfs_busy lock order

On Mon, Jun 04, 2018 at 01:12:40AM +0200, Alexander Bluhm wrote: > dounmount() does the vfs busy in the forward direction of the mnt_list. > while ((mp = TAILQ_NEXT(mp, mnt_list)) != NULL) { > error = vfs_busy(mp, VB_WRITE|VB_WAIT); > Then it unmounts all nested moun

Towards per-socket locks: sofree() & sounlock()

Diff below adds a 'struct socket *' argument to sounlock() in order to prepare the stack for per-socket locks. That means sofree() will now unlock a given socket before freeing it. But since we do not want to not release the NET_LOCK() when processing incoming TCP packets, in_pcbdetach() needs a s

Fewer sofree()

For pfkey and routing sockets, calling sofree() in pr_detach is a noop. The reason is that `SS_NOFDREF' hasn't been set at this point so the socket won't be freed. So I'd like to remove the sofree() and add an assert instead. sofree() will need to change soon to be able to deal with per-socket lo

ed Remove BACKWARDS from put_tty_line

Hello tech@, This should be the final patch to remove BACKWARDS from our ed. If ed were compiled without the BACKWARDS flag it would have a pager option when running with the list command. This would trigger if a line is long enough to wrap enough times to fill all the rows of the output device.

Unlock sendit() based syscalls

Diff below contains the interesting bits to unlock most of the network related syscalls. As previously explained [0], we know that `f_data' is immutable for sockets so we only have to protect `f_count'. This is done by using a global mutex: `fhdlk'. I'm aware that this is not the best solution,

Re: [patch] Add kvm_close in mib_hrsystemprocs function

On Thu, 31 May 2018 17:40:36 +0800 Nan Xiao wrote: > Hi Gerhard, > > Thanks for your reply! > > Yes, if no "kvm_close(kd);", there will be resource (memory, file > descriptor) leak. So hope you can commit it, thanks! > > > On 5/30/2018 4:49 PM, Gerhard Roth wrote: > > On Wed, 30 May 2018 16:25