1There is a potential for deadlock when allocating a struct sk_buff for 2data that needs to be written out to aoe storage. If the data is 3being written from a dirty page in order to free that page, and if 4there are no other pages available, then deadlock may occur when a 5free page is needed for the sk_buff allocation. This situation has 6not been observed, but it would be nice to eliminate any potential for 7deadlock under memory pressure. 8 9Because ATA over Ethernet is not fragmented by the kernel's IP code, 10the destructor member of the struct sk_buff is available to the aoe 11driver. By using a mempool for allocating all but the first few 12sk_buffs, and by registering a destructor, we should be able to 13efficiently allocate sk_buffs without introducing any potential for 14deadlock. 15