[9fans] p9p vac troubles, possible fix and questions?

[9fans] p9p vac troubles, possible fix and questions?

[Venkatesh Srinivas]

I believe that the bug that people have been having with vac and venti
in p9p (create bsize 8192 psize 8160vac: vacfscreate: vacfileroot:
read too small: asked for 0 need at least 389) can be fixed by this
change:

(in plan9port/src/libventi/fcall.c:185), change "f->count = (buf[2] <<
8) | buf[3];"  to "f->count =  (buf[0] << 8) | buf[1];"

I checked packet dumps of what vac was sending and what libventi was
reconstructing; I have the packet traces saved, if they'd be useful to
anyone.

As an aside, is Plan 9 slated to get this extended Venti protocol
(version 4) too? Is there any reason to support larger blocks? Was
there any reason venti limited at 56kb blocks in the past?

Also, has anyone ever been able to debug p9p apps with gdb? I try to
set breakpoints, but the apps just blow right through them, I'm not
sure why... if not, is there any other answer for debugging p9p apps?

Take care,
-- vs

Re: [9fans] p9p vac troubles, possible fix and questions?

[david bulkow]

[-- Attachment #1: Type: text/plain, Size: 318 bytes --]

>
> Also, has anyone ever been able to debug p9p apps with gdb? I try to
> set breakpoints, but the apps just blow right through them, I'm not
> sure why... if not, is there any other answer for debugging p9p apps?
>

For starters try setting:

(gdb) set follow-fork-mode child

before running the program.

[-- Attachment #2: Type: text/html, Size: 511 bytes --]

Re: [9fans] p9p vac troubles, possible fix and questions?

[Russ Cox]

Fixed in http://code.swtch.com/plan9port/changeset/c9f799b3ad09/
Thanks.