let pool allocators advertise multiple page sizes

this turns the pa_pagesz member of a pool allocator into a bitfield. pool pages can be many different sizes, they arent restricted to just the size provided by the hardware. to support this without introducing a page allocator per page size, a single page allocator can be used to allocate many dif

unifdef INSECURE

=== RCS file: /cvs/src/sys/stand/boot/cmd.c,v retrieving revision 1.63 diff -u -p -r1.63 cmd.c --- cmd.c 20 Jul 2014 19:33:54 - 1.63 +++ cmd.c 7 Nov 2016 22:47:40 - @@ -95,9 +95,7 @@ getcmd(void) int read_con

Re: The end of nd6_output()

Hi, On Mon, Nov 07, 2016 at 09:12:12AM +0100, Martin Pieuchot wrote: > [...] > Could you capture the route changes via "# route monitor"? I'd like > to know if the 'bad gateway value' message is triggered by userland > or the kernel. > [...] I can't reproduce the problem anymore. If it does turn

switchd(8): add more packet-out validations

Now that we have the flow-mod validation with the action/instructions support we can extend the usage of this functions for the packet-out validation. This diff increases the packet-out validation coverage by also doing instructions and packet truncation checks. ok? Index: ofp13.c ==

Re: Kill ifa_ifwithnet()

On Mon, 7 Nov 2016 08:59:53 +0100 Martin Pieuchot wrote: > On 04/11/16(Fri) 21:33, Vincent Gross wrote: > > [...] > > Why are you killing Strict Source Route Record ? Just as you did > > with rtredirect(), you can check whether RTF_GATEWAY is set and > > send back an ICMP_UNREACH if so. Or did I

Recursive splsoftnet() in PMTU

Timers configured with rt_timer_add(9) are always run under splsoftnet() so no need to take it recursively. ok? Index: netinet/ip_icmp.c === RCS file: /cvs/src/sys/netinet/ip_icmp.c,v retrieving revision 1.152 diff -u -p -r1.152 ip_i

Avoid spl recursion in doaccept()

We're aiming to replace critical sections protected by splsoftnet() by a non recursive rwlock. So we'll have to care about recursivity. Diff below prevents a recursion in the error path. Currently closef() will call soclose() which will take splsoftnet() again. So let's release the spl level be

splnet() in socket layer

splnet() was necessary when link state changes were executed from hardware interrupt handlers. In 2013 they got deferred to their own task, so KERNEL_LOCK() is what really protect these data structures. This is part of my bigger socket lock diff. ok? Index: kern/uipc_socket.c ==

Re: The end of nd6_output()

On 05/11/16(Sat) 19:26, Gregor Best wrote: > [...] > If I can do anything else to be of assistance, please let me know. Could you capture the route changes via "# route monitor"? I'd like to know if the 'bad gateway value' message is triggered by userland or the kernel.

Re: Stop watchdog timer during DDB session

On 02/11/16(Wed) 15:18, Christian Ludwig wrote: > Deliberately breaking into DDB, e.g. via serial console BREAK, can lead > to an abrupt end of the debugging session once the hardware watchdog > reboots the machine. This is because all CPUs are IPL_HIGH. None of them > tickles the watchdog anymore.

Re: Kill ifa_ifwithnet()

On 04/11/16(Fri) 21:33, Vincent Gross wrote: > [...] > Why are you killing Strict Source Route Record ? Just as you did with > rtredirect(), you can check whether RTF_GATEWAY is set and send back > an ICMP_UNREACH if so. Or did I miss something ? Like that? Index: netinet/ip_input.c