(moving from AIDO thread)
Regarding that aido-server and btdownloaddir takes 100% CPU.
The poll loop looks like this:
| gettimeofday | poll([{fd=3, events=POLLIN|POLLRDNORM}, {fd=11, events=POLLIN|POLLRDNORM}, {fd=15, events=POLLIN|POLLRDNORM}, {fd=17, events=POLLIN|POLLRDNORM}, {fd=139, events=POLLIN|POLLRDNORM}, {fd=36, events=POLLIN|POLLRDNORM}, {fd=104, events=POLLIN|POLLRDNORM}, {fd=53, events=POLLIN|POLLRDNORM}, {fd=117, events=POLLIN|POLLRDNORM}, {fd=119, events=POLLIN|POLLRDNORM}, {fd=132, events=POLLIN|POLLRDNORM}, {fd=120, events=POLLIN|POLLRDNORM}, {fd=16, events=POLLPRI|POLLRDBAND, revents=POLLHUP}, {fd=20, events=POLLIN|POLLRDNORM}, {fd=114, events=POLLIN|POLLRDNORM}, {fd=28, events=POLLIN|POLLRDNORM}, {fd=25, events=POLLIN|POLLRDNORM}, {fd=30, events=POLLIN|POLLRDNORM}, {fd=115, events=POLLIN|POLLRDNORM}], 19, 31) = 1 | gettimeofday({1083055259, 644064}, NULL) = 0 | gettimeofday({1083055259, 644102}, NULL) = 0 | gettimeofday({1083055259, 644238}, NULL) = 0 | poll
ad nauseum. Trace loop looks like this:
| - master.pike:2666: 82d5058->`()(3600.000000) | - master.pike:2666: 82d5058->`()(3600.000000) | - master.pike:2666: 82d5058->`()(3600.000000) | - master.pike:2666: 82d5058->`()(3600.000000)
fd=16 has revent=POLLHUP. Is that ignored by Pike? It doesn't look very good...
fd=16 has revent=POLLHUP. Is that ignored by Pike? It doesn't look very good...
Agreed. Which version of Pike?
As far as I can see, you should get a error, read or write callback on the fd informing you that the fd should be closed.
/ Henrik Grubbström (Lysator)
Previous text:
2004-04-27 10:44: Subject: 100%
(moving from AIDO thread)
Regarding that aido-server and btdownloaddir takes 100% CPU.
The poll loop looks like this:
| gettimeofday | poll([{fd=3, events=POLLIN|POLLRDNORM}, {fd=11, events=POLLIN|POLLRDNORM}, {fd=15, events=POLLIN|POLLRDNORM}, {fd=17, events=POLLIN|POLLRDNORM}, {fd=139, events=POLLIN|POLLRDNORM}, {fd=36, events=POLLIN|POLLRDNORM}, {fd=104, events=POLLIN|POLLRDNORM}, {fd=53, events=POLLIN|POLLRDNORM}, {fd=117, events=POLLIN|POLLRDNORM}, {fd=119, events=POLLIN|POLLRDNORM}, {fd=132, events=POLLIN|POLLRDNORM}, {fd=120, events=POLLIN|POLLRDNORM}, {fd=16, events=POLLPRI|POLLRDBAND, revents=POLLHUP}, {fd=20, events=POLLIN|POLLRDNORM}, {fd=114, events=POLLIN|POLLRDNORM}, {fd=28, events=POLLIN|POLLRDNORM}, {fd=25, events=POLLIN|POLLRDNORM}, {fd=30, events=POLLIN|POLLRDNORM}, {fd=115, events=POLLIN|POLLRDNORM}], 19, 31) = 1 | gettimeofday({1083055259, 644064}, NULL) = 0 | gettimeofday({1083055259, 644102}, NULL) = 0 | gettimeofday({1083055259, 644238}, NULL) = 0 | poll
ad nauseum. Trace loop looks like this:
| - master.pike:2666: 82d5058->`()(3600.000000) | - master.pike:2666: 82d5058->`()(3600.000000) | - master.pike:2666: 82d5058->`()(3600.000000) | - master.pike:2666: 82d5058->`()(3600.000000)
fd=16 has revent=POLLHUP. Is that ignored by Pike? It doesn't look very good...
/ Mirar
And if no callback is configured?
/ Per Hedbor ()
Previous text:
2004-04-27 11:21: Subject: 100%
fd=16 has revent=POLLHUP. Is that ignored by Pike? It doesn't look very good...
Agreed. Which version of Pike?
As far as I can see, you should get a error, read or write callback on the fd informing you that the fd should be closed.
/ Henrik Grubbström (Lysator)
CVS version of 7.5 as of Sunday or so.
I didn't get the same effect in 7.5.15.
/ Mirar
Previous text:
2004-04-27 11:21: Subject: 100%
fd=16 has revent=POLLHUP. Is that ignored by Pike? It doesn't look very good...
Agreed. Which version of Pike?
As far as I can see, you should get a error, read or write callback on the fd informing you that the fd should be closed.
/ Henrik Grubbström (Lysator)
pike-devel@lists.lysator.liu.se