Hey, today, I have run smatch and it reports, that aac_convert_sgraw2() returns -1 instead of -ENOMEM on kmalloc(). This is easy to fix, but nobody tests what aac_convert_sgraw2() returns in aac_build_sgraw2()
Since commit 0b4334473d48 ("[SCSI] aacraid: SCSI dma mapping failure
case handling"), aac_build_sgraw2() should have better error propagation
in aac_{read,write}_raw_io().
The question is, that if it's intentional (and I hope it's not) or it's
just leftover fix?
--
Nikola
