On Thu, 9 Apr 2015, Dan Carpenter wrote: > Sorry, my last email was bad. > > Splitting patches into logical parts is a bit tricky. Let me try > explain better. > > Every patch should sort of make sense on its own. In the original code > it's using GFP_ATOMIC but that's because the original API was bad and > we had no choice. In the 1/1 patch we're using GFP_ATOMIC explicitly > by choice and it's wrong. In patch 2/2 we fix this problem but we > shouldn't introduce bad code even if we fix it in later patches. But if Quentin's analysis is wrong, then we have to undo the GFP_KERNEL choice, and with only one patch we end up back at the pci API? julia -- To unsubscribe from this list: send the line "unsubscribe kernel-janitors" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
