Em Sat, Nov 17, 2007 at 03:42:52PM +0000, Gerrit Renker escreveu:
> Test Tree Inventory
> ===================
>
> This is the list of current small batches (from oldest to latest) as of Sat 17 Nov.
> The entire test tree is fully bisectable, so batches can be combined freely.
>
> Spaces in between sets indicate separate blocks.
>
> The CCID3 set has been completely re-arranged and recombined to give self-contained and
> independent patches. Within the set there are further, self-contained blocks.
>
>
> 1. Miscellaneous CCID3 fixes
> ----------------------------
> A small set of various fixes, independent of the main CCID3 patch set.
>
> [CCID3]: Revert use of MSS instead of s
Applied.
> [CCID3]: Ignore trivial amounts of elapsed time
Applied, but I think we can revisit elapsed time later by taking a
timestamp earlier, at dccp_v4_rcv, because then there can be time spent
in sk_backlog, etc. We could then store a timestamp on skb->cb and the
ccids would set a flag on dccp_sock to tell that the timestamp is
wanted.
Ah, and there is more space to save, after this patch the layout is
this, on x86_64:
struct ccid3_hc_rx_sock {
struct tfrc_rx_info ccid3hcrx_tfrc; /* 0 12 */
/* XXX 4 bytes hole, try to pack */
u64 ccid3hcrx_seqno_nonloss:48; /* 16 8 */
u64 ccid3hcrx_ccval_nonloss:4; /* 16 8 */
u64 ccid3hcrx_ccval_last_counter:4; /* 16 8 */
enum ccid3_hc_rx_states ccid3hcrx_state:8; /* 20 4 */
u32 ccid3hcrx_bytes_recv; /* 24 4 */
/* XXX 4 bytes hole, try to pack */
ktime_t ccid3hcrx_tstamp_last_feedback; /* 32 8 */
ktime_t ccid3hcrx_tstamp_last_ack; /* 40 8 */
struct list_head ccid3hcrx_hist; /* 48 16 */
/* --- cacheline 1 boundary (64 bytes) --- */
struct list_head ccid3hcrx_li_hist; /* 64 16 */
u16 ccid3hcrx_s; /* 80 2 */
/* XXX 2 bytes hole, try to pack */
u32 ccid3hcrx_pinv; /* 84 4 */
/* size: 88, cachelines: 2 */
/* sum members: 78, holes: 3, sum holes: 10 */
/* last cacheline: 24 bytes */
};
Just moving ccid3hcrx_bytes_recv to after ccid3hcrx_tfrc we can save 8
bytes.
> [CCID3]: Accurately determine idle & application-limited periods
Applied.
> [CCID3]: Inline for moving average
Applied
> 2. Main CCID3 patch set
> -----------------------
> This is the original CCID3 patch set, developed in Feb/Mar, significantly reordered to
> make the test tree fully bisectable.
> The patch set does three things (reflected in the order of batches):
>
> (1) new TX history for TFRC (packet_history.{c,h}),
Why do we need DECLARE_TFRC_TX_CACHE?
> (2) new RX history for TFRC (packet_history.{c,h}),
> (3) new Loss Intervals database (loss_interval.{c,h}).
>
> (Actually it is 3+1/2 things, the third patch introduces tfrc_module.c).
>
> [TFRC]: Migrate TX history to singly-linked list
> [TFRC]: Remove old (doubly-linked list) TX history
>
> [TFRC]: Provide central source file and debug facility
>
> [TFRC]: New RX history implementation
+static struct kmem_cache *tfrcxh;
Why "xh"?
> [CCID3]: Hook up with new RX history interface
> [TFRC]: Remove old RX history interface
Will try to continue later today.
- Arnaldo
-
To unsubscribe from this list: send the line "unsubscribe dccp" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
