RE: [t13] READ VERIFY SECTOR(S) (EXT) and NV cache command set

[Mark Overby]

Title: READ VERIFY SECTOR(S) (EXT) and NV cache command set

I think there is a lack of clarity.   The current READ AND VERIFY commands do not specify what happens if data is in the cache when it is verified. It is implied by the statement "The device shall read the data from the media and verify that there are no errors". This implies that if there is more current data in the cache then on the media, then the device shall write the data to the media and then execute the verify operation.   In corresponding logic, SCSI has the concept of a non-volatile cache. It explicitly states:   When a VERIFY command or a WRITE AND VERIFY command is processed, both a force unit access and a synchronize cache operation are implied, since the logical blocks are being verified as being stored on the medium. Therefore, I would argue that we need to be explict in either the READ VEIRFY SECTORS commands what happen or in the NV Cache command set. I prefer the former because it is the closest to the command.   My recommendataion would be that if a device has an LBA in the NV cache and a READ VERIFY SECTOR(S) (EXT) command is executed against that LBA, the device shall commit the data from the NV cache to media and verify the data was written correctly to the media.   I'll bring a proposal to the August meeting.   More ideas or feedback is welcome. From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Curtis Stevens Sent: Wednesday, July 12, 2006 11:39 AM To: forum@t13.org Subject: RE: [t13] READ VERIFY SECTOR(S) (EXT) and NV cache command set My guess is that this should be handled the same way it is handled today   Write cache is the same, regardless of the NV nature or not.  I am beginning to fail to see the issue.       ------------------------------------------------- Curtis E. Stevens 20511 Lake Forest Drive #C-214D Lake Forest, California 92630 Phone: 949-672-7933 Cell: 949-307-5050 E-Mail: [EMAIL PROTECTED]   From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Thomas Kjørnes Sent: Wednesday, July 12, 2006 10:31 AM To: 'Joseph Chen'; 'Mark Overby'; forum@t13.org Subject: RE: [t13] READ VERIFY SECTOR(S) (EXT) and NV cache command set   Wouldn’t it be best to verify both locations?   Regards, Thomas.   From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Joseph Chen Sent: 12. juli 2006 19:09 To: 'Mark Overby'; forum@t13.org Subject: RE: [t13] READ VERIFY SECTOR(S) (EXT) and NV cache command set   Mark,   When the data had been placed on two locations we face issue of the consistency. The data from the media maybe difference from the NV due to some potential conditions: 1. Disc data is broken from grown defects. 2. NV data is broken due to flash grown defects.   One of the purpose for the READ VERIFY command, as it was used in some “format” verification, is to find defects and map out those potential area from system file table (FAT or MFT). When the host sends READ VERIFY command, the device does a read from the media, and validates the ECC result to decide if the data is good. However, many of the devices implement bad sector reassignment so the OS defect mapping usage becoming unnecessary.   It will be better to define a consistent location to verify instead of either or.   Regards, Joseph   From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Mark Overby Sent: Tuesday, July 11, 2006 11:03 AM To: forum@t13.org Subject: [t13] READ VERIFY SECTOR(S) (EXT) and NV cache command set   If you issue as READ VERIFY SECTOR(S) (EXT) to an LBA that is pinned in the NV cache, should it verify the cache, the media, or both? This email message is for the sole use of the intended recipient(s) and may contain confidential information.  Any unauthorized review, use, disclosure or distribution is prohibited.  If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message.