On Mon, Apr 16, 2012 at 12:34:28AM +0200, Juan Francisco Cantero Hurtado wrote:
> On Sun, Apr 15, 2012 at 07:30:21PM +0200, Ariane van der Steldt wrote:
> > The reaper is not very nice: it blocks any access to the kernel (i.e.
> > syscalls) until it is done with its work. For big processes, this m
Hi,
This diff is called 'vmmap_reaperyield.diff.1_flags_only'. It's the
proper way forward for the reaperyield diff I mailed earlier.
This largely mechanical diff adds flags to uvm_unmap, uvmspace_free and
related functions. The diff should have no functional change. I need
this to make uvm_un
On Sun, Apr 15, 2012 at 07:30:21PM +0200, Ariane van der Steldt wrote:
> Hi,
>
> The reaper is not very nice: it blocks any access to the kernel (i.e.
> syscalls) until it is done with its work. For big processes, this may
> result in stalls in userspace. This diff attempts to relieve it by
> yi
Hi,
The reaper is not very nice: it blocks any access to the kernel (i.e.
syscalls) until it is done with its work. For big processes, this may
result in stalls in userspace. This diff attempts to relieve it by
yielding the reaper during this operation.
Please test this diff (vmmap_reaperyield.
> some software needs fully functional per-thread %gs base address,
Which software?
