crashes with more than 4 erupters

[udovdh]

When I start bfgminer like:

./bfgminer -S /dev/ttyUSB0 -S /dev/ttyUSB1 -S /dev/ttyUSB2 -S /dev/ttyUSB3 -S /dev/ttyUSB4 -S /dev/ttyUSB5 --icarus-timing short

it crashes soon after starting.

When I reduce the amount of erupters on the commandline to four it succeeds.

[udovdh]

./bfgminer -S all has a similar effect. Starting with four does allow to add erupters via the menu in most cases.

[udovdh]

(Fedora 19 on x86_64 with kernel.org 3.10.10)

[udovdh]

It even crashes with 4 erupters. 3.11.1 kernel. I do have s strace.

It ends like:

clock_gettime(CLOCK_MONOTONIC_RAW, {116, 567388958}) = 0 poll([{fd=29, events=POLLIN|POLLPRI|POLLRDNORM|POLLRDBAND}], 1, 0) = 0 (Timeout) sendto(29, "POST / HTTP/1.1\r\nAuthorization: "..., 299, MSG_NOSIGNAL, NULL, 0) = 299 poll([{fd=29, events=POLLIN|POLLPRI|POLLRDNORM|POLLRDBAND}, {fd=29, events=POLLOUT|POLLWRNORM}], 2, 0) = 1 ([{fd=29, revents=POLLOUT|POLLWRNORM}]) sendto(29, "{\"method\": \"getwork\", \"params\": "..., 44, MSG_NOSIGNAL, NULL, 0) = 44 poll([{fd=29, events=POLLIN}, {fd=29, events=POLLOUT}], 2, 1000) = 1 ([{fd=29, revents=POLLOUT}]) poll([{fd=29, events=POLLIN|POLLPRI|POLLRDNORM|POLLRDBAND}, {fd=29, events=POLLOUT|POLLWRNORM}], 2, 0) = 1 ([{fd=29, revents=POLLOUT|POLLWRNORM}]) poll([{fd=29, events=POLLIN}], 1, 1000) = 1 ([{fd=29, revents=POLLIN}]) poll([{fd=29, events=POLLIN|POLLPRI|POLLRDNORM|POLLRDBAND}], 1, 0) = 1 ([{fd=29, revents=POLLIN|POLLRDNORM}]) recvfrom(29, "HTTP/1.1 200 OK\r\nServer: nginx/1"..., 16384, 0, NULL, NULL) = 1026 clock_gettime(CLOCK_MONOTONIC_RAW, {117, 78804913}) = 0 clock_gettime(CLOCK_MONOTONIC_RAW, {117, 78857730}) = 0 sched_yield() = 0 clock_gettime(CLOCK_MONOTONIC_RAW, {117, 78979820}) = 0 sched_yield() = 0 futex(0x1103b3c, FUTEX_CMP_REQUEUE_PRIVATE, 1, 2147483647, 0x1103b10, 84) = 2 futex(0x1103b10, FUTEX_WAKE_PRIVATE, 1) = 1 sched_yield() = 0 clock_gettime(CLOCK_MONOTONIC_RAW, {117, 79440170}) = 0 sched_yield() = 0 sched_yield() = 0 sched_yield() = 0 sched_yield() = 0 sched_yield() = 0 sched_yield() = 0 sched_yield() = 0 sched_yield() = 0 sched_yield() = 0 clock_gettime(CLOCK_MONOTONIC_RAW, {117, 80123102}) = 0 poll([{fd=29, events=POLLIN|POLLPRI|POLLRDNORM|POLLRDBAND}], 1, 0) = 0 (Timeout) sendto(29, "POST / HTTP/1.1\r\nAuthorization: "..., 299, MSG_NOSIGNAL, NULL, 0) = 299 poll([{fd=29, events=POLLIN|POLLPRI|POLLRDNORM|POLLRDBAND}, {fd=29, events=POLLOUT|POLLWRNORM}], 2, 0) = 1 ([{fd=29, revents=POLLOUT|POLLWRNORM}]) sendto(29, "{\"method\": \"getwork\", \"params\": "..., 44, MSG_NOSIGNAL, NULL, 0) = 44 poll([{fd=29, events=POLLIN}, {fd=29, events=POLLOUT}], 2, 1000) = 1 ([{fd=29, revents=POLLOUT}]) poll([{fd=29, events=POLLIN|POLLPRI|POLLRDNORM|POLLRDBAND}, {fd=29, events=POLLOUT|POLLWRNORM}], 2, 0) = 1 ([{fd=29, revents=POLLOUT|POLLWRNORM}]) poll([{fd=29, events=POLLIN}], 1, 1000) = 1 ([{fd=29, revents=POLLIN}]) poll([{fd=29, events=POLLIN|POLLPRI|POLLRDNORM|POLLRDBAND}], 1, 0) = 1 ([{fd=29, revents=POLLIN|POLLRDNORM}]) recvfrom(29, "HTTP/1.1 200 OK\r\nServer: nginx/1"..., 16384, 0, NULL, NULL) = 1026 clock_gettime(CLOCK_MONOTONIC_RAW, {117, 599123337}) = 0 clock_gettime(CLOCK_MONOTONIC_RAW, {117, 599198543}) = 0 sched_yield() = 0 clock_gettime(CLOCK_MONOTONIC_RAW, {117, 599352439}) = 0 sched_yield() = 0 futex(0x1103b3c, FUTEX_CMP_REQUEUE_PRIVATE, 1, 2147483647, 0x1103b10, 92) = 2 futex(0x1103b10, FUTEX_WAKE_PRIVATE, 1) = 1 sched_yield() = 0 clock_gettime(CLOCK_MONOTONIC_RAW, <unfinished ...> +++ killed by SIGSEGV +++

If you need the complete strace let me know.

[luke-jr]

strace isn't useful for this kind of crash. I believe I may have already fixed it in git. Can you confirm?

[udovdh]

I just stopped bfgminer and restarted it specifying 6 erupters on the commandline. worked OK. So that indicates a change. (normally it would crash once or twice at least) Maybe some more testing to be 101% sure. Thanks for the git commit!

[Quix0r]

Just a question: @udovdh have you provided enough electricity? Every miner consumes a considerable amount and not all USB hubs are cable of handling it. So they "under-support" the devices which causes them to not work correctly (as they don't get the required ampere amount).

[udovdh]

There is plenty of power via two powered Konig 7-port hubs. The problem was fixed in software if I recall.

[Quix0r]

Ah, just to be sure. :)