On Thu, Aug 30, 2012 at 6:43 PM, Kent Overstreet [off-list ref] wrote:

On Thu, Aug 30, 2012 at 06:07:45PM -0400, Vivek Goyal wrote:

quoted

On Wed, Aug 29, 2012 at 10:13:45AM -0700, Kent Overstreet wrote:

[..]

quoted

quoted

Performance aside, punting submission to per device worker in case of deep
stack usage sounds cleaner solution to me.

Agreed, but performance tends to matter in the real world. And either
way the tricky bits are going to be confined to a few functions, so I
don't think it matters that much.

If someone wants to code up the workqueue version and test it, they're
more than welcome...

Here is one quick and dirty proof of concept patch. It checks for stack
depth and if remaining space is less than 20% of stack size, then it
defers the bio submission to per queue worker.

I can't think of any correctness issues. I see some stuff that could be
simplified (blk_drain_deferred_bios() is redundant, just make it a
wrapper around blk_deffered_bio_work()).

Still skeptical about the performance impact, though - frankly, on some
of the hardware I've been running bcache on this would be a visible
performance regression - probably double digit percentages but I'd have
to benchmark it.  That kind of of hardware/usage is not normal today,
but I've put a lot of work into performance and I don't want to make
things worse without good reason.

Here's another crazy idea - we don't really need another thread, just
more stack space.

We could check if we're running out of stack space, then if we are
just allocate another two pages and memcpy the struct thread_info
over.

I think the main obstacle is that we'd need some per arch code for
mucking with the stack pointer. And it'd break backtraces, but that's
fixable.