On 6/20/22 1:45 AM, Christoph Hellwig wrote: > On Sun, Jun 19, 2022 at 11:25:33AM -0500, michael.christie@xxxxxxxxxx wrote: >> sg_write_same allows it. We found the bug because some user just decided >> to do: >> >> sg_write_same ... -nbod .. /dev/sdb >> >> and it crashed the box. > > Oh. > >> I didn't know about the MI_REPORT_SUPPORTED_OPERATION_CODES part of it. >> I don't need support for the feature. I just want to fix the crash. >> I prefer just returning failure since nothing ever has ever used it if >> other people prefer that as well. > > I think the feature is generally useful, and I know Martin had patches > to use it in Linux. But I think a minimal fix for the remotely I'll work with Martin to find if there is an oracle user to test and on a longer term feature addition. > exploitable crash has the highest priority. Where does it crash? It crashes when we first access the sg in file and iblock's execute_write_same functions. > Maybe we just need a better sanity check somewhere if a command > claims to transfer data but has not payload? I'll look into it and send a patch.
