Re: Squid 3.0 Compilation issues on IA64

Hello Henrik,

Thanks for your response. Looks like the problem is more than what I thought
off. Please see the details below.

I fixed all the warnings, most of them are from debug statements and just
commented them out. Other warnings and fixes are as follows...

1. "HttpReply.cc" - typecast to ssize_t solves this warning

161 size_t s = XMIN(end + 1, (ssize_t)4096);
                                                   ^^^^^^^
HttpReply.cc: In function `int httpReplyParse (HttpReply *, const char *,
long int)':
HttpReply.cc:161: no matching function for call to `min (long int, int)'

2. "mem.cc" - %d to %ld

cc1plus: warnings being treated as errors
mem.cc: In function `void memBufStats (StoreEntry *)':
mem.cc:125: warning: int format, different type arg (arg 3)
mem.cc:125: warning: int format, different type arg (arg 4)
mem.cc: In function `void Mem::PoolReport (const MemPoolStats *, const
MemPoolMeter *, StoreEntry *)':
mem.cc:622: warning: int format, different type arg (arg 4)
mem.cc:622: warning: int format, different type arg (arg 5)
mem.cc:622: warning: int format, different type arg (arg 9)
mem.cc:622: warning: int format, different type arg (arg 10)
mem.cc:622: warning: int format, different type arg (arg 13)
mem.cc:622: warning: int format, different type arg (arg 14)

storeAppendPrintf(sentry, "Large buffers: %ld (%ld KB)\n",
    124 HugeBufCountMeter.level,
    125 HugeBufVolumeMeter.level / 1024);

    598 storeAppendPrintf(e,
    599 "%d\t %ld\t %ld\t %.2f\t %.1f\t" /* alloc */
    600 "%d\t %ld\t %ld\t %.1f\t" /* in use */
    601 "%d\t %ld\t %ld\t" /* idle */
    602 "%.0f\t %.1f\t %.1f\t %.1f\n", /* saved */
    603 /* alloc */
    604 mp_st->items_alloc,

3. "stat.c" - %6d to %6ld

cc1plus: warnings being treated as errors
stat.cc: In function `void info_get (StoreEntry *)':
stat.cc:691: warning: int format, different type arg (arg 3)
stat.cc:693: warning: int format, different type arg (arg 3)

    690 storeAppendPrintf(sentry, "\tmemPool accounted: %6ld KB
%3d%%\n",
    691 mp_stats.TheMeter->alloc.level >> 10,
percent(mp_stats.TheMet er->alloc.level, t));
    692 storeAppendPrintf(sentry, "\tmemPool unaccounted: %6ld KB
%3d%%\n",
    693 (t - mp_stats.TheMeter->alloc.level) >>
10, percent((t - mp_s tats.TheMeter->alloc.level), t));

> > Finally, I am NOT quite happy about the hack I did, but works fine.
> > I did some quick testing with my netscape. Looks better except the
> > above log messages. May be I should fix all the warnings, instead
> > of removing "-Werror" flag from the compilation. If I come up with
> > some changes to make the above issue vanished, will the changes be
> > acceptable to go into CVS?
>
> The warnings should be fixed.

Please see the above details. As I stated already, other warnings are from
debug statements. I can send them, if you need.

But this[Warning fixes] does NOT solve the runtime issue, I have reported
earlier. Still, the assertion is getting failed in the same location.

2003/05/25 16:44:08| Starting Squid Cache version 3.0.DEVEL-20030524 for
ia64-unknown-linux-gnu...
2003/05/25 16:44:08| assertion failed: mem.cc:230: "size ==
StrPoolsAttrs[i].obj_size"
Aborted

Any help will be much appreciated to proceed further.

> > Another weired thing I have seen is, regardless of the value I have
> > set for "mem_pools" "on" or "off", I always have seen "mem_pools
> > are off" in the log file as follows. I am NOT quite sure about
> > this, just wanted to know, if there is a known issue...
> > Log info : Memory pools are 'off'; limit: 0.00 MB
>
> And you did not request to have the pools disabled when you built your
> Squid? (--disable-mempools)

I have used only default option with my own prefix. By default, mempools
should be on, I guess.

Regards
GV

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
www.ViSolve.net - The Open Source Solutions Provider
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
Received on Sun May 25 2003 - 17:52:26 MDT