On Tue, Feb 05, 2019 at 07:28:41PM +0900, Tetsuo Handa wrote: > ashmem_pin() is calling range_shrink() without checking whether > range_alloc() succeeded. Since memory allocation fault injection might > force range_alloc() to fail while range_alloc() is called for only once > for one ioctl() request, make range_alloc() not to fail.
Why does this not need to fail? I am worried your change will introduce unwanted endless looping in the kernel instead of gracefully failing a pin/unpin request. Unless there is a good reason, I suggest to drop this patch from the series; but let us discuss more if you want. thanks, - Joel >From the docs: __GFP_NOFAIL - Indicate that this allocation is in no way allowed to fail (think twice before using). _______________________________________________ devel mailing list de...@linuxdriverproject.org http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
