Re: 10.x performance issue?

[Mouse <mouse%Rodents-Montreal.ORG@localhost>]

> Someone who rejects emails wrote:

Yeah, I could snark at you, too.  But what's the point?

>>>> Problem is, starting X takes foreeeeever [under 10.1]
>> I've now actually tried this for myself.
>> I have a little more information.
> Maybe, maybe not.

Well, I have more information; it may not be _useful_... :-)

> There are multiple applications that get started together, you have
> only traced one of them.

Very good point.

I ran the whole command under ktrace -i, so I actually have ktrace data
for each piece.  I just checked the whole file, sorting by syscall
times, and I find

The `bad' run
   546    546 xinit    0.753117553 RET   __nanosleep50 0
   544    549 X        0.296895070 RET   poll 1
   544    544 X        0.243910223 RET   ioctl 0
   544    549 X        0.109929501 RET   poll 1
   544    549 X        0.039969679 RET   poll 1
   544    544 X        0.039901056 CALL  mmap(0,0xb000,PROT_READ|PROT_WRITE,0x1002<PRIVATE,ANONYMOUS,ALIGN=NONE>,0xffffffff,0,0)
   544    549 X        0.036890455 RET   poll 1
   544    549 X        0.036889406 RET   poll 1
   544    544 X        0.029997974 RET   poll 0
   544    544 X        0.029997673 RET   poll 0
   544    544 X        0.029996880 RET   poll 0
   544    544 X        0.029996210 RET   poll 0
   544    549 X        0.026892913 RET   poll 1
   544    544 X        0.019547689 RET   poll 0
   544    544 X        0.016640645 GIO   fd 14 read 32 bytes
   544    544 X        0.016640250 GIO   fd 14 read 32 bytes

The `good' run
   651    646 X        0.958075514 RET   poll 1
   652    652 xinit    0.749048999 RET   __nanosleep50 0
   651    651 X        0.244047365 RET   ioctl 0
   651    646 X        0.099944713 RET   poll 1
   651    646 X        0.048070375 RET   poll 1
   651    646 X        0.038074019 RET   poll 1
   651    651 X        0.029997741 RET   poll 0
   651    651 X        0.029996762 RET   poll 0
   651    651 X        0.029982132 RET   poll 0
   651    646 X        0.028070687 RET   poll 1
   651    651 X        0.021887630 RET   poll 0
   651    646 X        0.019969427 RET   poll 1
   651    646 X        0.018071397 RET   poll 1
   651    651 X        0.017139996 RET   poll 0
   651    651 X        0.014538550 RET   poll 0
   651    651 X        0.014473515 RET   poll 0

So at least there's no single syscall that's responsible.

> The only time that I see any delay in starting up X11 is if there is
> a networking configuration problem that would cause a DNS lookup to
> time out.

That's one of my go-to explanations whenever anything network-capable
is taking unusually long.

But, in this case, the only difference is the kernel.  So, while it's
not entirely excluded, I do consider it a less plausible explanation in
this case - the change has to follow from *something* kernel.

> My guess is that the DNS lookup is to setup server access control.

I don't entirely rule it out.  But I do consider it unlikely, because,
with the exact same userland (including X server and all clients),

(a) 9.1 kernel, without -depth 8: fast
(b) 9.1 kernel, with -depth 8: fast
(c) 10.1 kernel, without -depth 8: fast
(d) 10.1 kernel, with -depth 8: slow

I'm having trouble constructing a plausible scenario in which -depth 8
leads to more trouble resolving something in the DNS.

Maybe tech-x11 would be more useful?  I've been using tech-kern because
I saw this as a kernel change leading to a performance issue (in the
above list of scenarios, changing from (b) to (d)), but it's starting
to look to me more like an X server issue provoked by a kernel
difference.

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		mouse%rodents-montreal.org@localhost
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B