Re: modem works!, battery drains fast?

2001-01-03 Thread Doug McNaught
Another thing you can do, that trades off Unix semantics for battery
efficiency, is to mount all your filesystems with the 'noatime'
option.  If you don't do this, even reading a file that's in cache
will cause a disk write, to update the access time field of the
inode.  There isn't much on a typical workstation that depends on
'atime' being correct, so it's probably worth doing for laptops. 

I got this off linux-kernel a while back and hadn't seen it mentioned
yet, so I thought I'd pipe up.

-Doug



Re: modem works!, battery drains fast?

2001-01-03 Thread Doug McNaught
Michel Dänzer <[EMAIL PROTECTED]> writes:

> Doug McNaught wrote:
> > 
> > Another thing you can do, that trades off Unix semantics for battery
> > efficiency, is to mount all your filesystems with the 'noatime'
> > option.  If you don't do this, even reading a file that's in cache
> > will cause a disk write, to update the access time field of the
> > inode.  There isn't much on a typical workstation that depends on
> > 'atime' being correct, so it's probably worth doing for laptops.
> 
> What exactly is 'not much'? :)

I can't think of anything offhand.  Several people on l-k reported
using it with no problems, so it's worth a try. 

-Doug



Re: modem works!, battery drains fast?

2001-01-03 Thread Doug McNaught
Michel Dänzer <[EMAIL PROTECTED]> writes:

> Ethan Benson wrote:
> > 
> > does cron not check /etc/cron.d often?  i know you can drop files in
> > there and cron will use them without restarting it so it must check
> > this somehow...
> 
> According to the manpage it does so every minute, and I haven't found a way
> around this so it could be the showstopper.

If you mount with 'noatime', this should not cause a problem--cron.d
and its files will live in cache, and you'll never hit the disk.

-Doug



Re: debian: cvs server giving permission denied

2001-04-03 Thread Doug McNaught
Brendan J Simon <[EMAIL PROTECTED]> writes:

> I've got a CVS server on a PowerMac running Debian GNU Linux
> (stable).  I can checkout from the repository but I can not do
> simple operations like status or update.  This only happens when
> performing these operations on directories with subdirectories
> (excluding CVS).  Is this a known problem and how do I solve it.

Sanity check:  does the user the server runs as have write access to
all the directories in the repository?

-Doug



Bug#68804: mount(8) and losetup(8) do not support int. kernel (PATCH)

2000-08-08 Thread Doug McNaught
Package: mount
Version: 2.10f-5.1

Debian Woody for PowerPC.  Potato seems to share the same mount
package so this is a bug in 'frozen' FWIW.  Bug applies to all
architectures AFAICS.

The 'mount' and 'losetup' commands do not work properly with the
international kernel.  If I compile Serpent into the kernel, and do

# losetup -e serpent /dev/loop0 

I get "Unsupported encryption type serpent".

The first problem is that the binary package 2.10f-5.1 does not
support Serpent--doing a 'strings /sbin/losetup | grep serpent' 
returns nothing.  This is very odd since the source package for the
same version has Serpent support (it's added by the .diff.gz file). 
It appears that the PowerPC version of the package (at least) was
miscompiled. 

Once I started hacking around in the source, I found that support was
missing in the password prompt code for Serpent and several other
ciphers.  I ended up getting the latest util-linux patch from the
kerneli tree and applying it by hand (only the password stuff was
missing as it ended up).

The code as it's distributed also depends on an up-to-date
 to compile.  The version distributed with libc6
2.1.3-10 is missing #defines for Serpent and many other ciphers.  The
util-linux source includes its own version of loop.h for international 
support, but it's not used.

So here's the patch:

diff -ru util-linux-2.10f.orig/mount/Makefile util-linux-2.10f/mount/Makefile
--- util-linux-2.10f.orig/mount/MakefileTue Aug  8 22:03:31 2000
+++ util-linux-2.10f/mount/Makefile Tue Aug  8 22:29:32 2000
@@ -1,7 +1,7 @@
 include ../make_include
 include ../MCONFIG
 
-CFLAGS = -I$(LIB) $(OPT)
+CFLAGS = -I$(LIB) -Iint-headers $(OPT)
 WARNFLAGS = -Wall -Wstrict-prototypes -Wmissing-prototypes
 DEFINES = -DHAVE_NFS
 
diff -ru util-linux-2.10f.orig/mount/lomount.c util-linux-2.10f/mount/lomount.c
--- util-linux-2.10f.orig/mount/lomount.c   Tue Aug  8 22:03:31 2000
+++ util-linux-2.10f/mount/lomount.cTue Aug  8 22:37:21 2000
@@ -236,6 +236,10 @@
 
   case LO_CRYPT_IDEA:
   case LO_CRYPT_CAST128:
+  case LO_CRYPT_SERPENT:
+  case LO_CRYPT_MARS:
+  case LO_CRYPT_RC6:
+  case LO_CRYPT_DFC:
 pass = getpass("Password :");
 MDcalc((byte *)loopinfo.lo_encrypt_key,pass,strlen(pass));
 loopinfo.lo_encrypt_key_size=16; /* 128 Bit key */
diff -ru util-linux-2.10f.orig/mount/losetup.c util-linux-2.10f/mount/losetup.c
--- util-linux-2.10f.orig/mount/losetup.c   Tue Aug  8 22:03:31 2000
+++ util-linux-2.10f/mount/losetup.cTue Aug  8 22:39:42 2000
@@ -151,8 +151,12 @@
MDcalc((byte *)loopinfo.lo_encrypt_key,pass,strlen(pass));
loopinfo.lo_encrypt_key_size=20; /* 160 Bit key */
break;
-   case LO_CRYPT_CAST128:
case LO_CRYPT_IDEA:
+   case LO_CRYPT_CAST128:
+   case LO_CRYPT_SERPENT:
+   case LO_CRYPT_MARS:
+   case LO_CRYPT_RC6:
+   case LO_CRYPT_DFC:
pass = getpass("Passphrase :");
MDcalc((byte *)loopinfo.lo_encrypt_key,pass,strlen(pass));
loopinfo.lo_encrypt_key_size=16; /* 128 Bit key */



Re: can't play mp3s

2000-10-08 Thread Doug McNaught
Shad Gregory <[EMAIL PROTECTED]> writes:

> ** WARNING **: oss_open(): Failed to open audio device (/dev/dsp): No such 
> device

"No such device" means that the device entry is there but there is no
kernel driver attached to it.  Make sure you have compiled the driver
into the kernel or loaded the appropriate module.

-Doug



Re: Install problem No init found

2000-11-04 Thread Doug McNaught
[EMAIL PROTECTED] writes:

> BootX: Unperturbed I next installed BootX v1.2.2 into the System Folder using
> vmlinux as boot kernel. Things when a little better this time if only because
> I had text on screen. But the system still hung at this error message:
> 
> Warning: unable to open init console.  Kernel Panic: No init found. Try
> passing init= option to kernel.

Did you give BootX a ramdisk to use?  Otherwise the kernel isn't going 
to be able to do much.

-Doug