Re: [RFC 2/9] zfcp: decouple TMF response handler from scsi_cmnd

2017-08-01 Thread Steffen Maier
Just for the records, in case anyone wants to resurrect this later on: This patch is buggy. On 07/25/2017 04:14 PM, Steffen Maier wrote: Do not get scsi_device via req->data any more, but pass an optional(!) scsi_device to zfcp_fsf_fcp_handler_common(). The latter must now guard any access to

Re: [RFC 2/9] zfcp: decouple TMF response handler from scsi_cmnd

2017-07-25 Thread Hannes Reinecke
On 07/25/2017 04:14 PM, Steffen Maier wrote: > Do not get scsi_device via req->data any more, but pass an optional(!) > scsi_device to zfcp_fsf_fcp_handler_common(). The latter must now guard > any access to scsi_device as it can be NULL. > > Since we always have at least a zfcp port as scope,

[RFC 2/9] zfcp: decouple TMF response handler from scsi_cmnd

2017-07-25 Thread Steffen Maier
Do not get scsi_device via req->data any more, but pass an optional(!) scsi_device to zfcp_fsf_fcp_handler_common(). The latter must now guard any access to scsi_device as it can be NULL. Since we always have at least a zfcp port as scope, pass this as mandatory argument to