default search path for shared libraries

2020-03-22 Thread Jeffrey Walton 
Hi everyone,

I'm reading https://netbsd.gw.com/cgi-bin/man-cgi?ld.elf_so+1+NetBSD-current.
I'm trying to understand why a test suite is failing due to missing
shared objects.

Under LD_LIBRARY_PATH, the man page says:

A colon separated list of directories, overriding the
default search path for shared libraries.

What is the default search path? How is the list of paths formed?

The man page also says ld.so.conf(5) is used, but the ELF FAQ says it
is not. Also see https://www.netbsd.org/docs/elf.html.

Which is correct?

Thanks in advance.

Re: Pkgtools setup

2020-03-22 Thread Benny Siegert 
On Sun, Mar 22, 2020 at 3:02 PM Todd Gruhn  wrote:
> How much has pkgtools changed since 2010?

A lot.

> Are any vars set in /etc/profile?
>
> I have mk.conf copied to the /etc dir.

Look, if I were you, and I have weird inexplicable performance issues,
I would start with an empty /etc/mk.conf and no custom settings in the
profile. Then try again :)

-- 
Benny

Pkgtools setup

2020-03-22 Thread Todd Gruhn 
How much has pkgtools changed since 2010?
Are any vars set in /etc/profile?

I have mk.conf copied to the /etc dir.

Is /etc/ld.so.conf  used anymore?

Re: Tweaks to avoid ide timeouts under qemu?

2020-03-22 Thread Jaromír Doleček 
Hello,

can you please compile your kernel with ATADEBUG option, then for
running system with new kernel set wdcdebug_wd_mask variable to '2'
via ddb before the daily run, and then collect the output after it
hangs again? Beware it's going to be a lot of messages.

I'm particularly interested if the abort routine runs properly after
the DRQ timeout, and whether there is any spurious intr happening.

Jaromir

Le jeu. 19 mars 2020 à 02:26, Paul Ripke  a écrit :
>
> On Thu, Mar 12, 2020 at 12:53:47PM +1100, Paul Ripke wrote:
> > Just wondering what other people are doing to avoid ide timeouts when
> > running qemu? My disk image is on spinning rust, and when the /etc/daily
> > jobs run on the host, the qemu emulated netbsd kernel tends to see
> > timeouts and ultimately hang. Eg:
>
> Any ideas? Seems to happen pretty much every night if I leave it
> running.
>
> Mar 19 10:27:21 qemu /netbsd: [ 51459.2816107] autoconfiguration error: 
> piixide0:0:0: lost interrupt
> Mar 19 10:27:21 qemu /netbsd: [ 51459.2816107]  type: ata tc_bcount: 10752 
> tc_skip: 0
> Mar 19 10:27:21 qemu /netbsd: [ 51459.5670834] autoconfiguration error: 
> piixide0:0:0: bus-master DMA error: missing interrupt, status=0x21
> Mar 19 10:27:21 qemu /netbsd: [ 51459.5670834] wd0a: DMA error writing fsbn 
> 1801707 of 1801707-1801727 (wd0 bn 1801770; cn 879 tn 49 sn 10), xfer 38, 
> retry 0
> Mar 19 10:27:21 qemu /netbsd: [ 51480.5385488] autoconfiguration error: 
> piixide0:0:0: lost interrupt
> Mar 19 10:27:21 qemu /netbsd: [ 51480.5385488]  type: ata tc_bcount: 10752 
> tc_skip: 0
> Mar 19 10:27:21 qemu /netbsd: [ 51480.5385488] autoconfiguration error: 
> piixide0:0:0: bus-master DMA error: missing interrupt, status=0x21
> Mar 19 10:27:21 qemu /netbsd: [ 51481.4286980] wd0a: DMA error writing fsbn 
> 1801707 of 1801707-1801727 (wd0 bn 1801770; cn 879 tn 49 sn 10), xfer 38, 
> retry 1
> Mar 19 10:27:21 qemu /netbsd: [ 51490.7125714] wd0: soft error (corrected) 
> xfer 38
> Mar 19 10:27:21 qemu /netbsd: [ 51530.4227159] autoconfiguration error: 
> piixide0:0:0: lost interrupt
> Mar 19 10:27:21 qemu /netbsd: [ 51530.4227159]  type: ata tc_bcount: 16384 
> tc_skip: 0
> Mar 19 10:27:21 qemu /netbsd: [ 51530.4227159] autoconfiguration error: 
> piixide0:0:0: bus-master DMA error: missing interrupt, status=0x21
> Mar 19 10:27:21 qemu /netbsd: [ 51531.1769776] wd0: autoconfiguration error: 
> transfer error, downgrading to PIO mode 4
> Mar 19 10:27:21 qemu /netbsd: [ 51531.1769776] wd0(piixide0:0:0): using PIO 
> mode 4
> Mar 19 10:27:21 qemu /netbsd: [ 51531.1870047] wd0a: DMA error writing fsbn 
> 1801899 of 1801899-1801930 (wd0 bn 1801962; cn 879 tn 55 sn 10), xfer 38, 
> retry 0
> Mar 19 10:27:21 qemu /netbsd: [ 51552.7631096] autoconfiguration error: 
> piixide0:0:0: lost interrupt
> Mar 19 10:27:21 qemu /netbsd: [ 51552.7631096]  type: ata tc_bcount: 16384 
> tc_skip: 0
> Mar 19 10:27:21 qemu /netbsd: [ 51552.7631096] wd0a: device timeout writing 
> fsbn 1801899 of 1801899-1801930 (wd0 bn 1801962; cn 879 tn 55 sn 10), xfer 
> 38, retry 1
> Mar 19 10:27:21 qemu /netbsd: [ 51556.9003752] wd0: soft error (corrected) 
> xfer 38
> Mar 19 10:27:21 qemu /netbsd: [ 51573.7527689] piixide0:0:0: not ready, 
> st=0xd0, err=0x00
> Mar 19 10:27:21 qemu /netbsd: [ 51573.7527689] wd0a: device timeout writing 
> fsbn 1801979 of 1801963-1801983 (wd0 bn 1802042; cn 879 tn 57 sn 26), xfer 
> 38, retry 0
> Mar 19 10:27:21 qemu /netbsd: [ 51585.2350924] piixide0:0:0: timeout waiting 
> for DRQ, st=0x50, err=0x00
> 
>
> --
> Paul Ripke
> "Great minds discuss ideas, average minds discuss events, small minds
>  discuss people."
> -- Disputed: Often attributed to Eleanor Roosevelt. 1948.

NetBSD iconv ebcdic failures?

2020-03-22 Thread Jeffrey Walton 
Hi Everyone,

I was testing libxml2 on NetBSD 8.1. libxml2 was failing about a 8 or
10 self tests. All appear to be related to ebcdic. Also see
https://mail.gnome.org/archives/xml/2020-March/msg4.html.

I think we isolated the problem to the NetBSD's iconv(3) library. When
I built libxml2 against a fresh copy of iconv and gettext, then the
failures went away. For completeness, that's libiconv-1.16.tar.gz and
gettext-0.20.1.tar.gz, which are the latest release tarballs.

Jeff