: Simplify control flow in the calculation of t_ipi
PATCH 2/6 : Simplify control flow in the calculation of t_ipi
[CCID 3]: Simplify control flow in the calculation of t_ipi
This patch performs a simplifying (performance) optimisation:
In each call of the inline function ccid3_calc_new_t_ipi(), the state is tested against TFRC_SSTATE_NO_FBACK. This is expensive when the function is called very often. A simpler solution, implemented by this patch, is to adapt the control flow.
Background:
-----------
Upon sender initialisation, the values of t_ipi, t_nom = t_0, delta, as well as the initial packet sending rate, are all constants [RFC 3448, 4.2]. Until feedback arrives, these values are not changed. Hence it is not necessary to recalculate t_ipi, t_nom, delta until the first feedback has arrived, i.e. as long as the state TFRC_SSTATE_NO_FBACK persists.
Justification:
--------------
ccid3_calc_new_t_ipi() is called only in two places:
* in ccid3_hc_tx_packet_recv(); here the state is never TFRC_SSTATE_NO_FBACK, due to
if (hctx->ccid3hctx_state = TFRC_SSTATE_NO_FBACK) {
ccid3_hc_tx_set_state(sk, TFRC_SSTATE_FBACK);
/* ... */ * in ccid3_hc_tx_packet_sent()
=> Only in the second case, the state might be TFRC_SSTATE_NO_FBACK and this only
during the initial phase of a connection.
Solution:
---------
This patch avoids such a recalculation by a simple change of control flow in ccid3_hc_tx_packet_sent(). As a consequence, ccid3_calc_new_t_ipi() is never called in the state TFRC_SSTATE_NO_FBACK, hence the test against this state can safely be removed.
In addition, a problematic comment in ccid3_calc_new_t_ipi() was removed:
* the first part of the comment (initial t_ipi = 1 second) is correct
* the second part of the comment is not correct wrt. [RFC 3448, 4.4]
--- net/dccp/ccids/ccid3.c | 2 ++ 1 file changed, 2 insertions(+)
net/dccp/ccids/ccid3.c +3 -7
@@ -103,13 +103,7 @@ static void ccid3_hc_tx_set_state(struct/* Calculate new t_ipi (inter packet interval) by t_ipi = s / X_inst */ static inline void ccid3_calc_new_t_ipi (struct ccid3_hc_tx_sock *hctx) { -/* -* If no feedback spec says t_ipi is 1 second (set elsewhere and then -* doubles after every no feedback timer (separate function) -*/ -if (hctx->ccid3hctx_state !=TFRC_SSTATE_NO_FBACK ) - hctx->ccid3hctx_t_ipi =usecs_div (hctx->ccid3hctx_s, - hctx->ccid3hctx_x); + hctx->ccid3hctx_t_ipi =usecs_div (hctx->ccid3hctx_s, hctx->ccid3hctx_x); }/* Calculate new delta by delta = min(t_ipi / 2, t_gran / 2) */
@@ -394,6 +388,8 @@ static void ccid3_hc_tx_packet_sent(stru "as a data packet", dccp_role(sk));return ;case TFRC_SSTATE_NO_FBACK : +/* t_nom, t_ipi, delta do not change until feedback arrives */ +return ;case TFRC_SSTATE_FBACK :if (len >0 ) { timeval_sub_usecs(&hctx->ccid3hctx_t_nom,
Quoted 37 lines · Gerrit Renker
[CCID 3]: Simplify control flow in the calculation of
t_ipiThis patch performs a simplifying (performance) optimisation:
In each call of the inline function
ccid3_calc_new_t_ipi(), the state is tested againstTFRC_SSTATE_NO_FBACK. This is expensive when the function is called very often. A simpler solution, implemented by this patch, is to adapt the control flow.Background:
-----------
Upon sender initialisation, the values oft_ipi,t_nom=t_0, delta, as well as the initial packet sending rate, are all constants [RFC 3448, 4.2]. Until feedback arrives, these values are not changed. Hence it is not necessary to recalculatet_ipi,t_nom, delta until the first feedback has arrived, i.e. as long as the stateTFRC_SSTATE_NO_FBACKpersists.Justification:
--------------
ccid3_calc_new_t_ipi()is called only in two places:* in ccid3_hc_tx_packet_recv(); here the state is never TFRC_SSTATE_NO_FBACK, due to if (hctx->ccid3hctx_state = TFRC_SSTATE_NO_FBACK) { ccid3_hc_tx_set_state(sk, TFRC_SSTATE_FBACK); /* ... */ * in ccid3_hc_tx_packet_sent()=> Only in the second case, the state might be
TFRC_SSTATE_NO_FBACKand this only
during the initial phase of a connection.Solution:
---------
This patch avoids such a recalculation by a simple change of control flow inccid3_hc_tx_packet_sent(). As a consequence,ccid3_calc_new_t_ipi()is never called in the stateTFRC_SSTATE_NO_FBACK, hence the test against this state can safely be removed.
Quoted 3 lines
In addition, a problematic comment in
ccid3_calc_new_t_ipi()was removed:
* the first part of the comment (initialt_ipi= 1 second) is correct
* the second part of the comment is not correct wrt. [RFC 3448, 4.4]
Why is the second part of the comment wrong?
RFC 3448, 4.4:
4.4. Expiration of nofeedback timer
If the nofeedback timer expires, the sender should perform the following actions:
1) Cut the sending rate in half. If the sender has received feedback from the receiver, this is done by modifying the sender's cached copy of X_recv (the receive rate). Because the sending rate is limited to at most twice X_recv, modifying X_recv limits the current sending rate, but allows the sender to slow-start, doubling its sending rate each RTT, if feedback messages resume reporting no losses.
If the sending rate is halved, doesn't it implies the inter packet interval is doubled?
Otherwise I'm fine with the patch, just waiting for comments above this specific part (comment removal).
- Arnaldo
Quoted 59 lines · Arnaldo Carvalho de Melo
Quoted 37 lines · Gerrit Renker
[CCID 3]: Simplify control flow in the calculation of
t_ipiThis patch performs a simplifying (performance) optimisation:
In each call of the inline function
ccid3_calc_new_t_ipi(), the state is tested againstTFRC_SSTATE_NO_FBACK. This is expensive when the function is called very often. A simpler solution, implemented by this patch, is to adapt the control flow.Background:
-----------
Upon sender initialisation, the values oft_ipi,t_nom=t_0, delta, as well as the initial packet sending rate, are all constants [RFC 3448, 4.2]. Until feedback arrives, these values are not changed. Hence it is not necessary to recalculatet_ipi,t_nom, delta until the first feedback has arrived, i.e. as long as the stateTFRC_SSTATE_NO_FBACKpersists.Justification:
--------------
ccid3_calc_new_t_ipi()is called only in two places:* in ccid3_hc_tx_packet_recv(); here the state is never TFRC_SSTATE_NO_FBACK, due to if (hctx->ccid3hctx_state = TFRC_SSTATE_NO_FBACK) { ccid3_hc_tx_set_state(sk, TFRC_SSTATE_FBACK); /* ... */ * in ccid3_hc_tx_packet_sent()=> Only in the second case, the state might be
TFRC_SSTATE_NO_FBACKand this only
during the initial phase of a connection.Solution:
---------
This patch avoids such a recalculation by a simple change of control flow inccid3_hc_tx_packet_sent(). As a consequence,ccid3_calc_new_t_ipi()is never called in the stateTFRC_SSTATE_NO_FBACK, hence the test against this state can safely be removed.Quoted 3 lines
In addition, a problematic comment in
ccid3_calc_new_t_ipi()was removed:
* the first part of the comment (initialt_ipi= 1 second) is correct
* the second part of the comment is not correct wrt. [RFC 3448, 4.4]Why is the second part of the comment wrong?
RFC 3448, 4.4:
4.4. Expiration of nofeedback timer
If the nofeedback timer expires, the sender should perform the following actions:
1) Cut the sending rate in half. If the sender has received feedback from the receiver, this is done by modifying the sender's cached copy of
X_recv(the receive rate). Because the sending rate is limited to at most twiceX_recv, modifyingX_recvlimits the current sending rate, but allows the sender to slow-start, doubling its sending rate each RTT, if feedback messages resume reporting no losses.
Quoted 5 lines
If the sending rate is halved, doesn't it implies the inter packet interval is doubled?
Otherwise I'm fine with the patch, just waiting for comments above this specific part (comment removal).
For now I'm merging the patch changing just the commit log not to state that the comment was wrong, i.e. I think its just superfluous :-)
As I'm putting these csets in the 'ccid3' branch of my net-2.6.20 git tree if you disagree about any of the things I change today (Sunday) we can always go back and fix it before sending the whole lot to Dave.
- Arnaldo
| > [CCID 3]: Simplify control flow in the calculation of t_ipi <snip>
| > In addition, a problematic comment in ccid3_calc_new_t_ipi() was removed: | > * the first part of the comment (initial t_ipi = 1 second) is correct | > * the second part of the comment is not correct wrt. [RFC 3448, 4.4] | | Why is the second part of the comment wrong? | | RFC 3448, 4.4: | | 4.4. Expiration of nofeedback timer | | If the nofeedback timer expires, the sender should perform the | following actions: | | 1) Cut the sending rate in half. If the sender has received feedback | from the receiver, this is done by modifying the sender's cached | copy of X_recv (the receive rate). Because the sending rate is | limited to at most twice X_recv, modifying X_recv limits the | current sending rate, but allows the sender to slow-start, | doubling its sending rate each RTT, if feedback messages resume | reporting no losses. | | If the sending rate is halved, doesn't it implies the inter packet | interval is doubled?
Yes that is right. Ian, I apologise - you were correct in this comment. Will re-send an update.
Quoting Arnaldo Carvalho de Melo:
| For now I'm merging the patch changing just the commit log not to state that the | comment was wrong, i.e. I think its just superfluous :-) | | As I'm putting these csets in the 'ccid3' branch of my net-2.6.20 git | tree if you disagree about any of the things I change today (Sunday) | we can always go back and fix it before sending the whole lot to Dave.
If you haven't committed these (they are not in your tree yet), can you please consider the attached update where I have corrected commit message? Will put this online, too.
--------------------> Updated Patch <-------------------------------------- [CCID 3]: Simplify control flow in the calculation of t_ipi
This patch performs a simplifying (performance) optimisation:
In each call of the inline function ccid3_calc_new_t_ipi(), the state is tested against TFRC_SSTATE_NO_FBACK. This is expensive when the function is called very often. A simpler solution, implemented by this patch, is to adapt the control flow.
Background:
-----------
Upon sender initialisation, the values of t_ipi, t_nom = t_0, delta, as well as the initial packet sending rate, are all constants [RFC 3448, 4.2]. Until feedback arrives, these values are not changed. Hence it is not necessary to recalculate t_ipi, t_nom, delta until the first feedback has arrived, i.e. as long as the state TFRC_SSTATE_NO_FBACK persists.
Justification:
--------------
ccid3_calc_new_t_ipi() is called only in two places:
* in ccid3_hc_tx_packet_recv(); here the state is never TFRC_SSTATE_NO_FBACK, due to
if (hctx->ccid3hctx_state = TFRC_SSTATE_NO_FBACK) {
ccid3_hc_tx_set_state(sk, TFRC_SSTATE_FBACK);
/* ... */ * in ccid3_hc_tx_packet_sent()
=> Only in the second case, the state might be TFRC_SSTATE_NO_FBACK and this only
during the initial phase of a connection.
Solution:
---------
This patch avoids such a recalculation by a simple change of control flow in ccid3_hc_tx_packet_sent(). As a consequence, ccid3_calc_new_t_ipi() is never called in the state TFRC_SSTATE_NO_FBACK, hence the test against this state can safely be removed. Additionally, a comment at the begin of this function is removed.
--- net/dccp/ccids/ccid3.c | 10 +++------- 1 file changed, 3 insertions(+), 7 deletions(-)
net/dccp/ccids/ccid3.c +3 -7
@@ -103,13 +103,7 @@ static void ccid3_hc_tx_set_state(struct/* Calculate new t_ipi (inter packet interval) by t_ipi = s / X_inst */ static inline void ccid3_calc_new_t_ipi (struct ccid3_hc_tx_sock *hctx) { -/* -* If no feedback spec says t_ipi is 1 second (set elsewhere and then -* doubles after every no feedback timer (separate function) -*/ -if (hctx->ccid3hctx_state !=TFRC_SSTATE_NO_FBACK ) - hctx->ccid3hctx_t_ipi =usecs_div (hctx->ccid3hctx_s, - hctx->ccid3hctx_x); + hctx->ccid3hctx_t_ipi =usecs_div (hctx->ccid3hctx_s, hctx->ccid3hctx_x); }/* Calculate new delta by delta = min(t_ipi / 2, t_gran / 2) */
@@ -395,6 +389,8 @@ static void ccid3_hc_tx_packet_sent(stru "as a data packet", dccp_role(sk));return ;case TFRC_SSTATE_NO_FBACK : +/* t_nom, t_ipi, delta do not change until feedback arrives */ +return ;case TFRC_SSTATE_FBACK :if (len >0 ) { timeval_sub_usecs(&hctx->ccid3hctx_t_nom,
Quoted 8 lines · Gerrit Renker
Quoting Arnaldo Carvalho de Melo:
| For now I'm merging the patch changing just the commit log not to state that the | comment was wrong, i.e. I think its just superfluous :-) | | As I'm putting these csets in the 'ccid3' branch of my net-2.6.20 git | tree if you disagree about any of the things I change today (Sunday) | we can always go back and fix it before sending the whole lot to Dave.If you haven't committed these (they are not in your tree yet), can you please
Its in the 'ccid3' branch, look at the bottoom of:
http://www.kernel.org/git/?p=linux/kernel/git/acme/net-2.6.20.git;a=summary
Or directly in:
http://www.kernel.org/git/?p=linux/kernel/git/acme/net-2.6.20.git;a=shortlog;hÌid3
But as I said in one of my messages yesterday, this one was just a first stab at merging your ccid3 work, I'm now getting back to getting the ones where there was general agreement to submit to Dave, then look again at the others.
Quoted 1 line
consider the attached update where I have corrected commit message?
I'll use the updated one.
Quoted 1 line
Will put this online, too.
Thanks,
- Arnaldo
Quoted 1 line
If you haven't committed these (they are not in your tree yet)
| Its in the 'ccid3' branch, look at the bottoom of: | | http://www.kernel.org/git/?p=linux/kernel/git/acme/net-2.6.20.git;a=summary | | Or directly in: | | http://www.kernel.org/git/?p=linux/kernel/git/acme/net-2.6.20.git;a=shortlog;hÌid3
I should have used these urls ... %-)
I pulled your tree and it was not `in' and so was lead on the wrong track. Have bookmarked them - very handy to have.
Quoted 3 lines · Gerrit Renker
[CCID 3]: Simplify control flow in the calculation of
t_ipi
Signature · 7 lines
Ian McDonald Web: http://wand.net.nz/~iam4 Blog: http://imcdnzl.blogspot.com WAND Network Research Group Department of Computer Science University of Waikato New Zealand
Quoted 5 lines · Ian McDonald
Quoted 3 lines · Gerrit Renker
[CCID 3]: Simplify control flow in the calculation of
t_ipiAcked-by Ian McDonald <[email protected]>
Thanks, but this one was the only one you hadn't acked when I submitted latest batch to Dave, and he already pulled it, thanks anyway!
- Arnaldo