• Bug#263136: This issue might break applications

    From Thomas Schmidt@1:229/2 to All on Wed Aug 11 11:00:17 2004
    From: [email protected]

    * Adrian Bunk schrieb am 07.08.04, um 16:57 Uhr:
    On Fri, Aug 06, 2004 at 08:24:16PM -0600, Bdale Garbee wrote:
    Adrian Bunk <[email protected]> writes:

    severity 263136 grave

    Grave? Surely you jest.

    It's an issue that should definitely fixed before Debian 3.1 .

    Yes, this issue should be discussed, but i do not see a simply way to
    solve it.

    One reason for keeping major 250 in sarge is that it is allmost sure
    that sarge wont ship (at release-time) a kernel, which uses 212, and
    the dvb-driver-source-package (for 2.4-users) also uses 250. So
    just changing makedev to use would 212 break the device-generation for
    most users of sarge. (At least a long as they use a kernel <=2.6.7 or
    the dvb-driver-source-package in sarge.)

    Ok, there are also reasons to change it, because it is clear that some sarge-users would like to use a kernel >=2.6.8, but i do not see how
    this change can be made without breaking the dvb-device-generation for
    the other sarge-users.


    Note: I am the (co-)maintainer of the packages linuxtv-dvb,
    linuxtv-dvb-apps and vdr, which all rely on the correct
    dvb-device-generation, as they use "makedev dvb" in their postinst.

    There is also at least one other package, which uses it in its
    postinst: dvbtune


    Regards,
    Thomas

    --
    Thomas Schmidt
    :wq

    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v1.2.5 (GNU/Linux)

    iD8DBQFBGdmjc9+NqwoydlIRAlAGAKCkfBedmW4rLc5jS/NhX1fJ1O9JBgCaAuB/ xbemkzI2D9s1/QRZlSJxdzU=
    =BuoY
    -----END PGP SIGNATURE-----

    --- SoupGate-Win32 v1.05
    * Origin: you cannot sedate... all the things you hate (1:229/2)
  • From Adrian Bunk@1:229/2 to Thomas Schmidt on Wed Aug 11 13:20:08 2004
    From: [email protected]

    On Wed, Aug 11, 2004 at 10:32:35AM +0200, Thomas Schmidt wrote:
    * Adrian Bunk schrieb am 07.08.04, um 16:57 Uhr:
    On Fri, Aug 06, 2004 at 08:24:16PM -0600, Bdale Garbee wrote:
    Adrian Bunk <[email protected]> writes:

    severity 263136 grave

    Grave? Surely you jest.

    It's an issue that should definitely fixed before Debian 3.1 .

    Yes, this issue should be discussed, but i do not see a simply way to
    solve it.

    One reason for keeping major 250 in sarge is that it is allmost sure
    that sarge wont ship (at release-time) a kernel, which uses 212, and
    the dvb-driver-source-package (for 2.4-users) also uses 250. So
    just changing makedev to use would 212 break the device-generation for
    most users of sarge. (At least a long as they use a kernel <=2.6.7 or
    the dvb-driver-source-package in sarge.)

    Ok, there are also reasons to change it, because it is clear that some sarge-users would like to use a kernel >=2.6.8, but i do not see how
    this change can be made without breaking the dvb-device-generation for
    the other sarge-users.
    ...

    I wouldn't call it nice, but what about the following workaround:

    Add two dvb targets to MAKEDEV, and add an init script to makedev that
    executes the correct one at boot time (depending on the kernel version).

    Regards,
    Thomas

    cu
    Adrian

    --

    "Is there not promise of rain?" Ling Tan asked suddenly out
    of the darkness. There had been need of rain for many days.
    "Only a promise," Lao Er said.
    Pearl S. Buck - Dragon Seed



    --
    To UNSUBSCRIBE, email to [email protected]
    with a subject of "unsubscribe". Trouble? Contact [email protected]

    --- SoupGate-Win32 v1.05
    * Origin: you cannot sedate... all the things you hate (1:229/2)
  • From Thomas Schmidt@1:229/2 to All on Wed Aug 11 20:00:21 2004
    From: [email protected]

    * Adrian Bunk schrieb am 11.08.04, um 12:57 Uhr:
    I wouldn't call it nice, but what about the following workaround:

    Add two dvb targets to MAKEDEV, and add an init script to makedev that executes the correct one at boot time (depending on the kernel version).

    Yes, this could work, but i would extend/change it a little bit.

    I would do the check for the running kernel inside makedev, so that
    "MAKEDEV dvb" decides which major-version to use, depending on the
    running kernel. (kernel >= 2.6.8-rc3 -> char-major-212, otherwise 250)

    Then the init-script (which makedev allready uses) would just need to
    call "MAKEDEV dvb" (but only when /dev/dvb/ allready exists).


    Regards,
    Thomas

    --
    Thomas Schmidt
    [email protected]

    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v1.2.5 (GNU/Linux)

    iD8DBQFBGlnfc9+NqwoydlIRAoeyAKDEa+wyGFXEAsDD4RPecvbwJPUkrQCgyWBG pIsunPmw4DFEaqSJU93cIpQ=
    =5oOz
    -----END PGP SIGNATURE-----

    --- SoupGate-Win32 v1.05
    * Origin: you cannot sedate... all the things you hate (1:229/2)
  • From Adrian Bunk@1:229/2 to Thomas Schmidt on Wed Aug 11 22:10:08 2004
    From: [email protected]

    On Wed, Aug 11, 2004 at 07:39:44PM +0200, Thomas Schmidt wrote:
    * Adrian Bunk schrieb am 11.08.04, um 12:57 Uhr:
    I wouldn't call it nice, but what about the following workaround:

    Add two dvb targets to MAKEDEV, and add an init script to makedev that executes the correct one at boot time (depending on the kernel version).

    Yes, this could work, but i would extend/change it a little bit.

    I would do the check for the running kernel inside makedev, so that
    "MAKEDEV dvb" decides which major-version to use, depending on the
    running kernel. (kernel >= 2.6.8-rc3 -> char-major-212, otherwise 250)
    ...

    Agreed.

    But parsing will be simpler if it's simply checked for >= 2.6.8 (it
    should be enough if all final kernels are correctly supported).

    Regards,
    Thomas

    cu
    Adrian

    --

    "Is there not promise of rain?" Ling Tan asked suddenly out
    of the darkness. There had been need of rain for many days.
    "Only a promise," Lao Er said.
    Pearl S. Buck - Dragon Seed



    --
    To UNSUBSCRIBE, email to [email protected]
    with a subject of "unsubscribe". Trouble? Contact [email protected]

    --- SoupGate-Win32 v1.05
    * Origin: you cannot sedate... all the things you hate (1:229/2)
  • From Bdale Garbee@1:229/2 to Adrian Bunk on Thu Aug 12 23:40:12 2004
    From: [email protected]

    Adrian Bunk <[email protected]> writes:

    But parsing will be simpler if it's simply checked for >= 2.6.8 (it
    should be enough if all final kernels are correctly supported).

    Someone want to offer me a patch that implements this? It's not hard, but I'm working on other things today and it'll get done faster if someone hands me a patch.

    There's already code in the consoleonly target to do kernel version checking.

    Bdale


    --
    To UNSUBSCRIBE, email to [email protected]
    with a subject of "unsubscribe". Trouble? Contact [email protected]

    --- SoupGate-Win32 v1.05
    * Origin: you cannot sedate... all the things you hate (1:229/2)
  • From Thomas Schmidt@1:229/2 to All on Fri Aug 13 00:40:10 2004
    From: [email protected]

    --wRRV7LY7NUeQGEoC
    Content-Type: text/plain; charset=us-ascii
    Content-Disposition: inline
    Content-Transfer-Encoding: quoted-printable

    * Bdale Garbee schrieb am 12.08.04, um 23:22 Uhr:
    Someone want to offer me a patch that implements this? It's not hard, but I'm
    working on other things today and it'll get done faster if someone hands me a patch.

    There's already code in the consoleonly target to do kernel version checking.

    Ah, this was the part where i did not know how to realize this check,
    thanks for the hint. :)

    Well, i created a patch for this bug now, but i am not completely sure
    about the check of the $kern_rev3, because it will also output just 8
    on kernels like 2.6.8-pre2 (where char-major 250 was used).

    The second thing where i am not sure is if it is needed to check for
    devfs before calling makedev, and if the ckeck in my patch is right.

    See the attached file.


    Thomas

    --
    Thomas Schmidt
    [email protected]

    --wRRV7LY7NUeQGEoC
    Content-Type: text/plain; charset=us-ascii
    Content-Disposition: attachment; filename="263136.diff" Content-Transfer-Encoding: quoted-printable

    diff -Nurd makedev-2.3.1/debian/init.d makedev-2.3.1.new/debian/init.d
    --- makedev-2.3.1/debian/init.d 2004-08-13 00:08:20.000000000 +0200
    +++ makedev-2.3.1.new/debian/init.d 2004-08-13 00:10:38.000000000 +0200
    @@ -8,6 +8,20 @@
    case "$1" in
    start)
    test -L /dev/MAKEDEV || ln -fs /sbin/MAKEDEV /dev/MAKEDEV
    +
    + # create dvb-devices, if the dvb-directory allready exists, and
    + # no devfs is used (this a workaround for the changed major-number
    + # of the dvb-devices, introduced in kernel 2.6.8, to be able to
    + # switch between kernels <= 2.6.7 and >= 2.6.8, without manual
    + # recreation of the devices (MAKEDEV dvb checks the kernel-version
    + # and uses the correct major-number))
    +
    + if [ ! -e /dev/.devfsd ]; then
    + if [ -d /dev/dvb ]; then
    + cd /dev && ./MA
  • From Thomas Schmidt@1:229/2 to All on Fri Aug 13 11:00:27 2004
    From: [email protected]

    --h31gzZEtNLTqOjlF
    Content-Type: text/plain; charset=us-ascii
    Content-Disposition: inline
    Content-Transfer-Encoding: quoted-printable

    * Thomas Schmidt schrieb am 13.08.04, um 00:20 Uhr:
    See the attached file.

    Sorry, the patch was incomplete, it would match only on 2.6.x-kernels,
    here is the version that should work with 2.7.x and 3.x also. :)

    --
    Thomas Schmidt
    :wq

    --h31gzZEtNLTqOjlF
    Content-Type: text/plain; charset=us-ascii
    Content-Disposition: attachment; filename="263136.diff" Content-Transfer-Encoding: quoted-printable

    diff -Nurd makedev-2.3.1/debian/init.d makedev-2.3.1.new/debian/init.d
    --- makedev-2.3.1/debian/init.d 2004-08-13 00:08:20.000000000 +0200
    +++ makedev-2.3.1.new/debian/init.d 2004-08-13 00:10:38.000000000 +0200
    @@ -8,6 +8,20 @@
    case "$1" in
    start)
    test -L /dev/MAKEDEV || ln -fs /sbin/MAKEDEV /dev/MAKEDEV
    +
    + # create dvb-devices, if the dvb-directory allready exists, and
    + # no devfs is used (this a workaround for the changed major-number
    + # of the dvb-devices, introduced in kernel 2.6.8, to be able to
    + # switch between kernels <= 2.6.7 and >= 2.6.8, without manual
    + # recreation of the devices (MAKEDEV dvb checks the kernel-version
    + # and uses the correct major-number))
    +
    + if [ ! -e /dev/.devfsd ]; then
    + if [ -d /dev/dvb ]; then
    + cd /dev && ./MAKEDEV dvb
    + fi
    + fi
    +
    ;
  • From Bdale Garbee@1:229/2 to Adrian Bunk on Sat Aug 7 14:00:11 2004
    From: [email protected]

    Adrian Bunk <[email protected]> writes:

    severity 263136 grave

    Grave? Surely you jest.

    Bdale


    --
    To UNSUBSCRIBE, email to [email protected]
    with a subject of "unsubscribe". Trouble? Contact [email protected]

    --- SoupGate-Win32 v1.05
    * Origin: you cannot sedate... all the things you hate (1:229/2)
  • From Adrian Bunk@1:229/2 to All on Sat Aug 7 00:40:07 2004
    From: [email protected]

    severity 263136 grave
    thanks

    This issue definitely needs investigation before Debian 3.1 .

    cu
    Adrian

    --

    "Is there not promise of rain?" Ling Tan asked suddenly out
    of the darkness. There had been need of rain for many days.
    "Only a promise," Lao Er said.
    Pearl S. Buck - Dragon Seed




    --
    To UNSUBSCRIBE, email to [email protected]
    with a subject of "unsubscribe". Trouble? Contact [email protected]

    --- SoupGate-Win32 v1.05
    * Origin: you cannot sedate... all the things you hate (1:229/2)