286640 - performance optimizations in pk11wrap

Status: NEW

(NSS :: Libraries, defect, P2)

Description

[Julien Pierre]

After analyzing runtime code paths with dtrace, I came up with some quick and
dirty optimizations to accelerate pk11wrap for purposes of SSL benchmarking . I
will attach a set of patches for them. Most of those patches delete code that I
deemed unnecessary for benchmarking. They would need to be cleaned up to make
that code optional. They are not currently ready for checkin by any means, but I
wanted to have a record of them.

Comment 1

[Julien Pierre]

Attachment: don't try to ever login

Assumes token never logs itself out. Should be possible to configure if user
knows this for a fact.

Comment 2

[Julien Pierre]

Attachment: don't lock in PK11Context

This lock appears to be unnecessary for SSL purposes  . This is one of 4
PR_Lock operations that happens for every PKCS#11 hashing and encryption calls.
The other 3 are in the softoken.

Comment 3

[Julien Pierre]

Attachment: hardcode answer to certain slot lookups

This patch caches the slot used for 2 operations used frequently in SSL
handshakes. It's clearly wrong to do this in pk11wrap. The caching of slot
configuration should happen at a higher level, in libssl. See bug 285538 . But
this quick and dirty hack works when benchmarking SSL with NSS softoken .

Comment 4

[Robert Relyea]

These patches appear to be experiments in looking for bottle necks, and most
should (obviously) not go into the tree as is. I'm reassigning to Julien, and
adding myself to the CC list. If the experiments produce the need for additional
work, we can spin off additional bugs that refer back to this one.

bob

Comment 5

[Julien Pierre]

I'm reassigning this performance bug to Nelson .
I'm setting a target for 3.11, since this is our "performance release".
However, we are getting most our benefits in SSL from bypassing PKCS#11, except
for the RSA op. Because of this, it may be that some of these optimizations
don't need to get in after all - at least not urgently. Nelson, feel free to
lower the priority.

Comment 6

[Nelson Bolyard (seldom reads bugmail)]

remove target milestone, since the target was missed.

Comment 7

[BugBot [:suhaib / :marco/ :calixte]]

The bug assignee is inactive on Bugzilla, and this bug has priority 'P2'.
:beurdouche, could you have a look please?

For more information, please visit auto_nag documentation.

Comment 8

[Marco Castelluccio [:marco]]

We have modified the bot to only consider P1 as high priority, so I'm cancelling the needinfo here.