Oliver Hartkopp wrote:
> Hi Wolfgang,
> 
> i discovered a problem in the current mainline linux-2.6 git due to the commit
> of the EMS USB driver.
> 
> As the EMS USB driver has been (accidently?) added right after the EMS PCI
> driver the SJA1000 drivers had been torn in pieces and the indention is broken
> for CAN_KVASER_PCI.

I already realized this problem. The simplest fix is to move the the
Kconfig entry for the EMS USB driver below the SJA1000 ones.

> 
> I attached a (trivial) patch to move the SJA1000 and USB Kconfig files into
> their specific directories, which fixes the problem.

I like that fix even more.

> 
> But before we post a patch like this, the question comes up, if we should
> create directories for
> 
>   - CAN controllers (SJA1000, AT91, CC770, ...)
>   - Bus systems (USB, PCMCIA, IO, ...)
>   - Vendors (EMS, IXXAT, PEAK, homebrew, ...)

I don't like a separate directory per vendor.

> E.g. should the EMS PCMCIA driver appear in a pcmcia directory (and Kconfig)
> or is sja1000 still the better approach?

It should remain in sja1000 because it uses the sja1000dev interface
(CONFIG_CAN_SJA1000).

> For me it would be ok to sort the directory by CAN controllers, 'usb',
> 'softing', etc - as we have it right now.

The rule of thumb is: if there are more than *3* files for a driver or
class of drivers, a sub-directory could/should be created. For this
reason I was also happy with the EMS USB CAN driver in the top-level
directory. Have a look to "drivers/net":

  $ ls -1 drivers/net/fec_mpc52xx*
  drivers/net/fec_mpc52xx.c
  drivers/net/fec_mpc52xx.h
  drivers/net/fec_mpc52xx_phy.c

  $ ls -1 drivers/net/e1000/
  e1000_ethtool.c
  e1000.h
  e1000_hw.c
  e1000_hw.h
  e1000_main.c
  e1000_osdep.h
  e1000_param.c
  Makefile

> But at least we should move the Kconfigs into the subdirs as suggested in my
> patch.

Yes, this is less error prune. Are you going to provide a patch?

> Any other ideas?

Currently, I do not see a real need for moving files around.

Wolfgang.

_______________________________________________
Socketcan-core mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/socketcan-core

Reply via email to