Got it, thanks. Steve
-----Original Message----- From: Javen Wu [mailto:[EMAIL PROTECTED] Sent: Tuesday, April 08, 2008 6:55 PM To: Steve Chang Cc: 'ChaoHong Guo'; [email protected] Subject: Re: [driver-discuss] SCSI HBA driver debugging questions Yes., You shouldn't mount before format/label and newfs on it. mount is a filesystem operation, only why you make specific file system type on any slice, you mount it and operate it. Javen Steve Chang wrote: >Javen, >The disks are new. It's not been done any format >or label before. Shouldn't I mount it first before >making a new file system and labeling it? > > > >Steve > > >-----Original Message----- >From: Javen Wu [mailto:[EMAIL PROTECTED] >Sent: Tuesday, April 08, 2008 2:44 AM >To: Steve Chang >Cc: ChaoHong Guo; [email protected] >Subject: Re: [driver-discuss] SCSI HBA driver debugging questions > >Hi Stephen, > >I don't think mode sense page3 returning check condition is the root >cause of your mount failure. >And the SCSI_CAP_GEOMETRY in tran_getcap is not mandatory. >Mode page 3 and 4 are obsolete and a lot of new scsi devices don't >support mode page 3 and 4. And solaris would generate disk geometry >based on capacity data. > >Let me ask you simple questions firstly. >1. Did you label your disk by format? I mean did you run "type" >subcommand of format(1M) to label your disk? >2. Did you make file system by mkfs(1M) or newfs(1M) on your specific disk? > >Javen > >ChaoHong Guo wrote: > > > >>Steve Chang 写道: >> >> >> >>>Hi Javen, >>>Now I had a mount device issue after driver is installed >>> >>>#mount /dev/dsk/c4t0d0ds0 /mnt >>> >>>=> mount: I/O error >>> Mount: cannot mount /dev/dsk/c4t0d0s0 >>> >>>The following are traces >>> >>>0 - test unit ready - return OK >>>0 - test unit ready - return OK >>>0x1a - mode sense, page code=3, length 0x24 - return check >>>condition 8 - read 1 sector from LAB=0 - return OK >>>8 - read 1 sector from LBA=1 - return OK >>>0x28 - read 1 sector from LBA=9dfffe - return OK >>>0x28 - read 1 sector from LBA=9dffff - return OK >>>8 - read 1 sector from LBA=0 - return OK >>>8 - read 1 sector from LBA=1 - return OK >>> >>> >>>Can you advice me what could be wrong? >>> >>> >>> >>> >>pc 3 is used by sd driver to get physical geometry. If the scsi disk >>doesn't support it, >>maybe you should support virtual geometry in your HBA driver, so that >>sd(7d) driver >>can fallback to virtual geometry. >> >> >>You can try implementing SCSI_CAP_GEOMETRY >></source/s?defs=SCSI_CAP_GEOMETRY> in tran_getcap(). >> >>BTW, what HBA driver you are writing ? >> >> >>Good luck, >>-minskey >> >> >> >> >> >>>Steve >>> >>> >>>_______________________________________________ >>>driver-discuss mailing list >>>[email protected] >>>http://mail.opensolaris.org/mailman/listinfo/driver-discuss >>> >>> >>> >> >> > > > > _______________________________________________ driver-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/driver-discuss
