On 2020-08-27 18:04, Matthew Wilcox wrote:
On Wed, Aug 26, 2020 at 07:57:15PM +0530, Pradeep P V K wrote:
Fix this by protecting fuse_copy_pages() with fc->lock.
No. This is a spinlock and fuse_copy_pages() can allocate memory
with GFP_KERNEL. You need to enable more debugging on your test
system.
Thanks Matthew for the review and comments. I will address this in my
next patch set. BTW, can you please share your thoughts, if there is a
way to extract fuse_connection Data Structure from fuse_copy_state DS ?
This would really help to use fc->lock only on intended function
(fuse_ref_page()).
