On 8/23/22 03:16, Michael Wu wrote:
Yes... My patch did seem suspicious. Here's the scene about the block I/O error: Some card reader does not respond the command 'MEDIUM REMOVAL PREVENT' correctly, as a result, the host does not send subsequent cmd 'MEDIUM REMOVAL ALLOW'/'MEDIUM REMOVAL PREVENT' before/after sleep, which leads to a enumeration failure after system resume. I wonder, without changing the behavior of the device, is there's a better way to solve this? -- Modifying the scsi core should not be a good idea though :(
The above is not clear to me. My understanding is that "MEDIUM REMOVAL PREVENT" is a sense code instead of a SCSI command?
Thanks for your kindly notice. I looked around in the latest linux mainline repo, but could not find this code. Where can I get this 2002 version of scsi_lib.c? Thank you.
Please take a look at https://stackoverflow.com/questions/3264283/linux-kernel-historical-git-repository-with-full-history. That web page has instructions for how to configure a git repository such that history goes back before the time when Linus started using git.
Bart.
