On 8/13/23 18:19, Damien Le Moal wrote:

On 8/12/23 06:35, Bart Van Assche wrote:

quoted

Make the error handler call .eh_prepare_resubmit() before resubmitting

This reads like the eh_prepare_resubmit callback already exists. But you are
adding it. So you should state that.

Hi Damien,

I will rephrase the patch description.

quoted

+++ b/drivers/scsi/Makefile.kunit

@@ -0,0 +1 @@
+obj-$(CONFIG_SCSI_ERROR_TEST) += scsi_error_test.o

All the above kunit changes (and the test changes below) seem unrelated to what
the commit message describes. Should these be split into a different patch ?

Some people insist on including unit tests in the same patch as
the patch that introduces the code that is being tested. I can
move the unit test into a separate patch if that is preferred.

quoted

+	/*
+	 * Call .eh_prepare_resubmit for each range of commands with identical
+	 * ULD driver pointer.
+	 */
+	list_for_each_entry_safe(scmd, next, done_q, eh_entry) {
+		struct scsi_driver *uld = scsi_cmd_to_driver(scmd);
+		struct list_head *prev, uld_cmd_list;
+
+		while (&next->eh_entry != done_q &&
+		       scsi_cmd_to_driver(next) == uld)
+			next = list_next_entry(next, eh_entry);
+		if (!uld->eh_prepare_resubmit)
+			continue;
+		prev = scmd->eh_entry.prev;
+		list_cut_position(&uld_cmd_list, prev, next->eh_entry.prev);
+		uld->eh_prepare_resubmit(&uld_cmd_list);

Is it guaranteed that all uld implement eh_prepare_resubmit ?

That is not guaranteed. Hence the if (!uld->eh_prepare_resubmit)
test in the above loop.

Thanks,

Bart.