Re: PR/53524 CVS commit: src/sys/dev/ic

[Markus Kilbinger <mk%kilbi.de@localhost>]

The following reply was made to PR kern/53524; it has been noted by GNATS.

From: Markus Kilbinger <mk%kilbi.de@localhost>
To: gnats-bugs%netbsd.org@localhost, Nick Hudson <nick.hudson%gmx.co.uk@localhost>, martin%netbsd.org@localhost
Cc: 
Subject: Re: PR/53524 CVS commit: src/sys/dev/ic
Date: Thu, 22 Nov 2018 01:08:16 +0100

 It took some time to rebuild HEADs tool and GENERIC kernel from
 scratch on my cubietruck (!).
 
 Compared to
   http://nycdn.netbsd.org/pub/NetBSD-daily/HEAD/201811211020Z/evbarm-earmv7hf/binary/sets/kern-GENERIC.tgz
 
 my GENERIC kernel is about 400 kb larger:
 
   tar tvzf kern-GENERIC.tgz.1 | grep netbsd.ub
   -rw-r--r--  1 root     wheel    8473324 Nov 21 10:49 ./netbsd.ub
 
 vs.
 
   rw-r--r--  1 root  wheel  8939436 Nov 21 20:33
 /usr/NetBSD/HEAD/src/sys/arch/evbarm/compile/GENERIC/netbsd.ub
 
 Does that mean anything, yet?
 
  >   >    > To generate a diff of this commit:
  >    >    >  cvs rdiff -u -r1.69 -r1.70 src/sys/dev/ic/ahcisata_core.c
  >    >
  >    >  This change breaks booting on my 'bananapro' (evbarm/)earmv7hf; stuck
  >    >  before normal 'Copyright ...' message) which has a sata connector, but
  >    >  no device is connected in my case.
  >
  >    > Are you *sure* it is this change? The code in question only runs
  >   much later...
 
 I don't know, but I can trigger maybe another (real) problem with that change.
 
 > Can you compare kernel sizes with and without the change? I agree with
 > Martin that this code runs much later.
 
 'netbsd.ub's size does not change with and w/o this change.
 
  Also, does the problem occur with VERBOSE_INIT_ARM/EARLYCONS defined?
 
 I can test that later (tomorrow?) ...
 
  >   (BTW: I switched from SUNXI to GENERIC kernel some time ago, because
  >   the SUNXI kernel no longer worked properly).
 
  > How did it stop working?
 
 SUNXI kernels panic / stuck suddenly, so I switch over to GENERIC kernel.
 If it helps I can build an actual SUNXI kernel and try again.
 
 Regards,
 
 Markus.