[illumos-Developer] missing lwp_exit() in kcfpool_svc()

[Garrett D'Amore]

  Webrev with the fix is at 
http://mexico.purplecow.org/gdamore/webrev/kcf-fix/

I've also updated the bug with a test module that exercises this in the 
kernel.  So far with my tests, I don't see any problem with the fix 
(examination of the system with the mdb commands listed shows that the 
lwp counts are reasonably in sync... the system count is 9 and the kcf 
count is 8, but that's because the system count includes the master 
dispatcher thread, so the off-by-one is expected.)

   - Garrett

On 03/31/11 11:52 PM, Bryan Cantrill wrote:
>> Please do file a bug in illumos; you are welcome to either set yourself, or
>> set me, as the responsible engineer.
> Done; it's 875.  I screwed up the formatting of the description, but
> couldn't figure out how to edit it, so my apologies for the poor
> formatting...
>
>> As I'm the guy that hacked that code
>> together to get rid of the userland process, its my fault.  :-/  I'm not
>> surprised I screwed this up actually, its the first time I've had to do
>> anything with lwps.
> And it's actually quite unusual to have in-kernel LWPs that exit, so
> there aren't many models to look at...
>
>> Generating lots of kcf jobs can be tricky.  If you have a hardware crypto
>> card, you can do it with OpenSSL.  You can also trigger this generically
>> with IPsec.  Doing a lot of work with /dev/random *might* trigger it, but
>> I'd have to look to see.
> We don't completely understand why he kept hitting it, but we're
> working on that angle. ;)
>
> Thanks for the help, for fixing this, and for getting us an open KCF
> to begin with!
>
>          - Bryan
>
> _______________________________________________
> Developer mailing list
> Developer at lists.illumos.org
> http://lists.illumos.org/m/listinfo/developer