diff options
| author | Thomas Gschwantner <tharre3@gmail.com> | 2018-07-18 17:47:50 +0200 |
|---|---|---|
| committer | Jason A. Donenfeld <Jason@zx2c4.com> | 2018-07-18 19:48:17 +0200 |
| commit | 1bd1d9660edf224c7b056a5df0b2ee4e526214c1 (patch) | |
| tree | 47251ab6541a3bd77f4a602ba2c01d2462dcf008 /src/compat/compat.h | |
| parent | f230add201d4963d97a09dcd9223900711fd1085 (diff) | |
recieve: disable NAPI busy polling
This avoids adding one reference per peer to the napi_hash hashtable, as
normally done by netif_napi_add(). Since we potentially could have up to
2^20 peers this would make busy polling very slow globally.
This approach is preferable to having only a single napi struct because
we get one gro_list per peer, which means packets can be combined nicely
even if we have a large number of peers.
This is also done by gro_cells_init() in net/core/gro_cells.c .
Signed-off-by: Thomas Gschwantner <tharre3@gmail.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Diffstat (limited to 'src/compat/compat.h')
| -rw-r--r-- | src/compat/compat.h | 9 |
1 files changed, 9 insertions, 0 deletions
diff --git a/src/compat/compat.h b/src/compat/compat.h index 997ae39..7943ab8 100644 --- a/src/compat/compat.h +++ b/src/compat/compat.h @@ -601,6 +601,15 @@ static inline void *skb_put_data(struct sk_buff *skb, const void *data, unsigned #define napi_complete_done(n, work_done) napi_complete(n) #endif +#if LINUX_VERSION_CODE < KERNEL_VERSION(4, 5, 0) +#include <linux/netdevice.h> +/* NAPI_STATE_SCHED gets set by netif_napi_add anyway, so this is safe. + * Also, kernels without NAPI_STATE_NO_BUSY_POLL don't have a call to + * napi_hash_add inside of netif_napi_add. + */ +#define NAPI_STATE_NO_BUSY_POLL NAPI_STATE_SCHED +#endif + /* https://lkml.kernel.org/r/20170624021727.17835-1-Jason@zx2c4.com */ #if IS_ENABLED(CONFIG_NF_CONNTRACK) #include <linux/ip.h> |