From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Vasquez Subject: Re: [PATCH 2/2] qla2xxx: Code changes for qla data structure refactoring Date: Thu, 22 Jan 2009 08:13:38 -0800 Message-ID: <20090122161338.GA30580@plap4-2.local> References: <20090122105016.GA100061@sgi.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from avexch1.qlogic.com ([198.70.193.115]:31364 "EHLO avexch1.qlogic.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753548AbZAVQNk (ORCPT ); Thu, 22 Jan 2009 11:13:40 -0500 Content-Disposition: inline In-Reply-To: <20090122105016.GA100061@sgi.com> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Jeremy Higdon Cc: Anirban Chakraborty , "James.Bottomley@HansenPartnership.com" , "James.Smart@emulex.com" , "linux-scsi@vger.kernel.org" , Linux Driver , "mdr@sgi.com" On Thu, 22 Jan 2009, Jeremy Higdon wrote: > On Thu, Nov 06, 2008 at 10:40:51AM -0800, Anirban Chakraborty wrote: > > Following changes have been made: > > 1. Outstanding commands are based on a request queue, scsi_qla_host does not > > maintain it anymore. > > 2. start_scsi is accessed via isp_ops struct instead of direct invocation. > > 3. Interrupt registrations are done using response queue instead of device id. > > > > Thanks, > > Anirban > > > > Signed-off-by: Anirban Chakraborty > > Hello Anirban, > > We found a problem in this patch. > The problem is that in the function qla2x00_probe_one() in qla_os.c, > you should be calling qla2x00_config_dma_addressing() before calling > qla2x00_mem_alloc(), so that the DMA mask can be set properly. > > Now I notice that qla2x00_config_dma_addressing() uses the virtual > SCSI host, while qla2x00_mem_alloc() uses the physical host adapter > structure. I don't see why qla2x00_config_dma_addressing() couldn't > be rewritten to use the physical host adapter structure. But if not, > then it may need to be split into two parts -- one to set the > consistent dma mask and another to set the streaming dma mask. Jeremy, Thanks for finding this. I'll add the following patch into my patch-queue if their's no major objections... -- av --- diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c index f344b68..703d8a8 100644 --- a/drivers/scsi/qla2xxx/qla_os.c +++ b/drivers/scsi/qla2xxx/qla_os.c @@ -65,8 +65,6 @@ MODULE_PARM_DESC(ql2xextended_error_logging, static void qla2x00_free_device(scsi_qla_host_t *); -static void qla2x00_config_dma_addressing(scsi_qla_host_t *ha); - int ql2xfdmienable=1; module_param(ql2xfdmienable, int, S_IRUGO|S_IRUSR); MODULE_PARM_DESC(ql2xfdmienable, @@ -1242,9 +1240,8 @@ qla2x00_change_queue_type(struct scsi_device *sdev, int tag_type) * supported addressing method. */ static void -qla2x00_config_dma_addressing(scsi_qla_host_t *vha) +qla2x00_config_dma_addressing(struct qla_hw_data *ha) { - struct qla_hw_data *ha = vha->hw; /* Assume a 32bit DMA mask. */ ha->flags.enable_64bit_addressing = 0; @@ -1870,6 +1867,7 @@ qla2x00_probe_one(struct pci_dev *pdev, const struct pci_device_id *id) set_bit(0, (unsigned long *) ha->vp_idx_map); + qla2x00_config_dma_addressing(ha); ret = qla2x00_mem_alloc(ha, req_length, rsp_length, &req, &rsp); if (!ret) { qla_printk(KERN_WARNING, ha, @@ -1896,8 +1894,6 @@ qla2x00_probe_one(struct pci_dev *pdev, const struct pci_device_id *id) pci_set_drvdata(pdev, base_vha); - qla2x00_config_dma_addressing(base_vha); - host = base_vha->host; base_vha->req_ques[0] = req->id; host->can_queue = req->length + 128;