On Mon, 2017-08-14 at 21:34 +0200, Borislav Petkov wrote: > On Mon, Aug 14, 2017 at 07:02:15PM +0000, Kani, Toshimitsu wrote: > > I do not know how likely we see such case, but the code should be > > written according to the spec. > > Well, then you'll have to make ghes_edac_report_mem_error() > reentrant. Which doesn't look that hard as the only thing it really > needs from struct ghes_edac_pvt are those string buffers. I guess you > can try to do the simplest thing first and allocate them on the > stack.
ghes_edac_report_mem_error() is reentrant as it is now. I think the current code design of allocating mci & ghes_edac_pvt for each GHES source entry makes sense. edac_raw_mc_handle_error() also has the same expectation that the call is serialized per mci. Thanks, -Toshi
