The sysinit structure is supposed to be a complete list of all peered i2o adapters. The current driver structure issued the sysinit command after all adapters were discovered rather than re-issuing as each was added. The structure still must be global, but I will look into moving the instantiation. Sincerely -- Mark Salyzyn -----Original Message----- From: Christoph Hellwig [mailto:hch@xxxxxxxxxxxxx] Sent: Friday, August 19, 2005 10:59 AM To: Salyzyn, Mark Cc: linux-scsi Subject: Re: [PATCH] dpt_i2o pci_driver model On Wed, Aug 17, 2005 at 01:04:05PM -0400, Salyzyn, Mark wrote: > Christoph Hellwig [mailto:hch@xxxxxxxxxxxxx] writes: > > ... update the driver to use the pci_driver model ... > > The patch applies to the scsi-misc-2.6 git tree. The moving of > adpt_install_hba to the near bottom of the dpt_i2o.c file creates a > dirty context diff in that region which is hard to inspect from within > the patch. No coding style changes were made when moving content from > the original initialization code to the module_init and module_exit > functions to aid before and after comparisons for inspection; even > despite the fact C++ style comments are used in these sections. We > expect such style cleanup to occur either on any necessary follow-up > rewrites after basic functional approval, or as a follow-up cleanup > patch. This looks wrong to me. It's still walking a lot of driver-global hba lists in the init routine. All hba initialization should happen in the ->probe callback, and there should be very little global initialization. There shouldn't really be any global list of hbas. - : send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
