Re: [PATCH] scsi: arcmsr: fix NULL deref on dma_alloc_coherent() failure in arcmsr_alloc_xor_buffer()

[Ivy Lopez <skunkolee@gmail.com>]

Thanks for the review.

Now that I look at it, the commit message's claim about a reachable
secondary NULL deref via arcmsr_free_ccb_pool()/arcmsr_iop_confirm()
doesn't hold up: arcmsr_alloc_ccb_pool() failing sends arcmsr_probe()
straight to unmap_pci_region, bypassing the free_ccb_pool label
entirely, so that path isn't reachable here. The acb->xor_mega = 0
line is still harmless defensively, but my stated justification for
it was wrong. My apologies for the inaccuracy!

Separately, thanks for flagging the pre-existing DMA leak of
acb->dma_coherent (and the IO queue) when arcmsr_alloc_xor_buffer()
fails inside arcmsr_alloc_ccb_pool()! I'll follow up with a separate
patch addressing that.

ivy