headers
Hauke Fath | 6 Nov 2007 21:14
Picon

Re: /cdrom

At 13:45 Uhr -0500 29.10.2007, Eric Haszlakiewicz wrote:

[removable media]

>Any reason not to setup an amd config for these directories?  ("too much
>more complicated than /etc/fstab"?)

As a sample configuration for amd(8), that's a good idea.

But the amd specific requirements violate POLA too often for it to be
useful in the hands of the less savvy. There is no 'umount -f' equivalent
-- 'amq -u' is a mere recommendation to amd(8). umount(8)ing a device owned
by amd(8) will confuse it terminally. And depending on how the kernel maps
the media's partition table to a disklabel, you have to keep around several
entries for one type of media (like USB sticks).

At 13:46 Uhr +0200 27.10.2007, Hubert Feyrer wrote:
>For something like /media or another more general scheme, I'd suggest to
>defer this discussion until we get something like a HAL/dbus
>infrastructure, and then think again on this whole thing.
>
>For now, I'd like to stick to the KISS principle.

Seconded.

	hauke

--
"It's never straight up and down"     (DEVO)
(Continue reading)

Permalink | Reply | 0 comments |
headers
Jeremy C. Reed | 17 Nov 2007 07:10

sysinst questions

Using NetBSD 4_RC4 for i386.

1) On the screen to set up disklabels, what does this (57) mean?

 MB	  	 Cylinders
160	(57)	 3414

2) Why does it have /usr, /var, /home as 0 MB, 0 cyl, and 0 sectors. Its 
note above on display says usually use whole disk for / but why even show 
them? It is confusing to me as it seems like something configured them 
wrong?

3) Also my swap was 0 MB. I don't know why. I may have misunderstood a 
previous step. But I thought that should have been configured for me also.

4) How to exit from that disklabel screen?

I saw no choice that says exit or done.

I pressed Ctrl-C and killed sysinst (not what I meant to do).
Permalink | Reply | 1 comment |
headers
David Laight | 17 Nov 2007 09:51
Picon

Re: sysinst questions

On Sat, Nov 17, 2007 at 12:10:02AM -0600, Jeremy C. Reed wrote:
> Using NetBSD 4_RC4 for i386.
> 
> 1) On the screen to set up disklabels, what does this (57) mean?
> 
>  MB	  	 Cylinders
> 160	(57)	 3414

That should be the size when the free space is added to the partition.
In this case the free space seems to be negative.

> 2) Why does it have /usr, /var, /home as 0 MB, 0 cyl, and 0 sectors. Its 
> note above on display says usually use whole disk for / but why even show 
> them? It is confusing to me as it seems like something configured them 
> wrong?

Because you always get a chance to to see what sysinst is going to do,
and to make changes yourself.

> 3) Also my swap was 0 MB. I don't know why. I may have misunderstood a 
> previous step. But I thought that should have been configured for me also.

It would, but the disk seems to be smaller than the minimum size for
the root filesystem.

> 4) How to exit from that disklabel screen?

The bottom row of the menu - but that will say:
    'Reduce partition sizes by xxx MB (xxx sectors).'
(Continue reading)

Permalink | Reply | 0 comments |
headers
Tetsuya Isaki | 18 Nov 2007 08:58
Picon

loadfile_*() returns an incorrect errno?

In fdloadfile() (in sys/lib/libsa/loadfile.c),
loadfile_*() returns -1 and sets global variable errno,
if error happened.  But loadfile_*() returns -1 without
setting errno, if read() is short.

Therefore, fdloadfile() may return error code 0 (= success)
when read() in loadfile_*() is short.

I think that loadfile_*() should always set errno when
returns -1.  So, how about the attached patch?  But I'm 
not sure whether ESHORT is good.

P.S. May I do a style fix in sys/lib/libsa?
---
Tetsuya Isaki <isaki <at> par.odn.ne.jp / isaki <at> NetBSD.org>

Index: sys/lib/libsa/loadfile_aout.c
===================================================================
RCS file: /cvsroot/src/sys/lib/libsa/loadfile_aout.c,v
retrieving revision 1.9
diff -u -r1.9 loadfile_aout.c
--- sys/lib/libsa/loadfile_aout.c	5 Jun 2007 08:48:50 -0000	1.9
+++ sys/lib/libsa/loadfile_aout.c	10 Nov 2007 12:49:50 -0000
 <at>  <at>  -105,6 +105,7  <at>  <at> 
 	paddr_t offset = marks[MARK_START];
 	u_long magic = N_GETMAGIC(*x);
 	int sub;
+	ssize_t nr;

 	/* some ports dont use the offset */
(Continue reading)

Permalink | Reply | 2 comments |
headers
Martin Husemann | 18 Nov 2007 11:25
Picon

Re: loadfile_*() returns an incorrect errno?

On Sun, Nov 18, 2007 at 04:58:02PM +0900, Tetsuya Isaki wrote:
> I think that loadfile_*() should always set errno when
> returns -1.  So, how about the attached patch?

Looks good.

>  But I'm 
> not sure whether ESHORT is good.

EREAD? But it's just a name, don't worry too much about it.

> P.S. May I do a style fix in sys/lib/libsa?

Sure, why not?

Martin
Permalink | Reply | 1 comment |
headers
Masao Uebayashi | 22 Nov 2007 10:58
Picon
Gravatar

Building single boot image

I don't think there is a way to build the single install boot image
(e.g., distrib/i386/floppies/bootfloppy-com) from build.sh or the top
level Makefile.  Do you think having that feature is good?  Has anyone
tried to achieve this?

I guess David Young's CUWiN has similar thing, but its infrastructure
is not compatible with NetBSD's.

Masao
Permalink | Reply | 1 comment |
headers
Tetsuya Isaki | 23 Nov 2007 05:30
Picon

Re: loadfile_*() returns an incorrect errno?

Hi, Martin.

Thanks for reviewing.
I'll commit it.
---
Tetsuya Isaki <isaki <at> par.odn.ne.jp / isaki <at> NetBSD.org>

At Sun, 18 Nov 2007 11:25:03 +0100,
Martin Husemann wrote:
> 
> On Sun, Nov 18, 2007 at 04:58:02PM +0900, Tetsuya Isaki wrote:
> > I think that loadfile_*() should always set errno when
> > returns -1.  So, how about the attached patch?
> 
> Looks good.
> 
> >  But I'm 
> > not sure whether ESHORT is good.
> 
> EREAD? But it's just a name, don't worry too much about it.
> 
> > P.S. May I do a style fix in sys/lib/libsa?
> 
> Sure, why not?
> 
> Martin
Permalink | Reply | 0 comments |

Gmane