|
222af657
|
2013-10-14T16:51:21
|
|
Remove hashfast write thread for all writes can be done from scanwork without added complexity.
|
|
|
e9b67b50
|
2013-10-14T16:19:52
|
|
Pass the thr argument to the read and write threads for hashfast to have access to its data.
|
|
|
5cbe4be9
|
2013-10-14T15:13:30
|
|
Incorporate hf protocol header changes.
|
|
|
e06e495a
|
2013-10-14T15:07:32
|
|
Merge branch 'master' into hashfast
|
|
5c7b9f7e
|
2013-10-13T18:52:51
|
|
Merge pull request #503 from kanoi/master
Klondike - driver redesign + api message correction
|
|
|
07837072
|
2013-10-14T12:12:01
|
|
hexstr is too small in test_work_current
|
|
|
166200c0
|
2013-10-14T12:01:40
|
|
Windows uses errno for WSAETIMEDOUT
|
|
|
02a4f8df
|
2013-10-14T11:44:37
|
|
Convert the usb callback function to using cgsem_t timed waits to avoid race conditions with conditionals/mutexes.
|
|
|
c9c39ac1
|
2013-10-14T11:44:13
|
|
Give correct return code in cgsem_mswait
|
|
|
1d9f90b2
|
2013-10-14T11:39:15
|
|
Revert "Convert the usb callback function to use a cgsem_t using the timedwait code to avoid races with mutexes/conditionals."
This reverts commit 433465bcfcd8d9a2270cb27dd0ea933e9fefa829.
Not ready yet.
|
|
|
433465bc
|
2013-10-14T11:35:47
|
|
Convert the usb callback function to use a cgsem_t using the timedwait code to avoid races with mutexes/conditionals.
|
|
|
6e948d87
|
2013-10-14T11:34:08
|
|
Check for correct timeout error in cgsem_mswait
|
|
|
51f8107b
|
2013-10-14T11:27:04
|
|
Fix util.h exports for cgsem_mswait
|
|
|
c75ce853
|
2013-10-14T11:22:57
|
|
Implement a generic cgsem_mswait similar to sem_timedwait
|
|
|
634151ff
|
2013-10-14T10:06:37
|
|
Merge branch 'master' into hashfast
|
|
|
9ca22ed4
|
2013-10-14T10:05:49
|
|
Use the one LIBUSB_ERROR_TIMEOUT for cancelled transactions since this error is explicitly tested for in various drivers.
|
|
|
a7580f1f
|
2013-10-14T09:40:23
|
|
Do not use locking on usb callback function pthread signalling to prevent deadlock with libusb's own event lock.
|
|
|
8d1604a2
|
2013-10-14T09:40:23
|
|
Do not use locking on usb callback function pthread signalling to prevent deadlock with libusb's own event lock.
|
|
|
d2d7a746
|
2013-10-14T00:54:39
|
|
Merge branch 'master' into hashfast
|
|
|
280dc4bf
|
2013-10-14T00:54:05
|
|
Lock mutex before second pthread_cond_wait in usbutils to prevent a race.
|
|
|
3c7af2e6
|
2013-10-14T00:28:57
|
|
Join threads to close hashfast driver.
|
|
|
598b58c8
|
2013-10-14T00:18:23
|
|
Create basic read and write threads that will be used by hashfast driver.
|
|
|
93d1f93f
|
2013-10-13T22:34:16
|
|
Perform remainder of hashfast_detect_common setup.
|
|
|
cbfa78ea
|
2013-10-13T22:26:26
|
|
Device data will not be allocated in hashfast_detect_one_usb so no point freeing it.
|
|
|
49befae6
|
2013-10-13T22:25:48
|
|
Get remaining data from reset sequence on hashfast driver.
|
|
|
922b4d85
|
2013-10-13T21:55:14
|
|
Store the device data returned from hashfast header returned in reset
|
|
|
775f2040
|
2013-10-13T21:49:57
|
|
Check return code from hashfast_get_header in reset and fail when appropriate.
|
|
|
d7a907f2
|
2013-10-13T21:45:01
|
|
Timeout should not be a fatal error in hashfast_get_header
|
|
|
58dd80e4
|
2013-10-13T21:39:59
|
|
Implement a hashfast get_header function which reads till it finds a header preamble or times out.
|
|
|
4b9eb370
|
2013-10-13T20:28:57
|
|
Implement hashfast send header function and add relevant usb op codes to arrays, beginning reset sequence on hashfast detection.
|
|
|
e7a9aef6
|
2013-10-13T18:48:56
|
|
Add hf_protocol.h to cgminer sources.
|
|
|
176cae27
|
2013-10-13T18:42:33
|
|
Add basic definitions for hashfast device recognition.
|
|
|
df6873ef
|
2013-10-13T17:23:51
|
|
Remove old initialisation code from hashfast, use hf protocol library where possible and prepare for new driver model.
|
|
|
f1941db0
|
2013-10-13T17:11:16
|
|
Minor style changes.
|
|
|
4664350c
|
2013-10-13T17:06:01
|
|
Include hf protocol header.
|
|
|
6e2d12e1
|
2013-10-13T16:57:28
|
|
Remove now unused crc32 code from hashfast driver.
|
|
|
9b5c5349
|
2013-10-13T16:41:53
|
|
Merge branch 'master' into hashfast
Conflicts:
api.c
configure.ac
miner.h
usbutils.c
usbutils.h
|
|
|
84de52c1
|
2013-10-13T10:11:46
|
|
Use a write lock when performing any USB control transfers to prevent concurrent transfers.
|
|
|
c0690286
|
2013-10-12T17:44:28
|
|
Free a libusb transfer after we have finished using it to avoid a dereference in usb_control_transfer
|
|
|
578fabe0
|
2013-10-12T16:54:39
|
|
Do not perform bfi int patching for opencl1.2 or later.
|
|
|
84f642f5
|
2013-10-12T15:06:48
|
|
Although async transfers are meant to use heap memory, we never return before the transfer function has completed so stack memory will suffice for control transfers, fixing a memory leak in the process.
|
|
16bd0a0d
|
2013-10-12T11:45:43
|
|
klondike - correct/reverse min/max stats
|
|
|
da7223f1
|
2013-10-12T11:30:57
|
|
api incorrect message name
|
|
|
f1d77f17
|
2013-10-12T11:29:02
|
|
klondike - use a link list queue rather than a circular buffer - and add timing stats
|
|
|
56edabc6
|
2013-10-12T11:06:54
|
|
Use a timeout with usb handle events set to a nominal 200ms and wait for the polling thread to shut down before deinitialising libusb.
|
|
|
9f8023a9
|
2013-10-12T10:34:07
|
|
Use stack memory for hex used in stratum share submissions.
|
|
|
e0c90359
|
2013-10-12T09:56:48
|
|
Use stack memory in test_work_current, avoiding a malloc/free cycle each time.
|
|
|
d3c215fd
|
2013-10-12T09:53:21
|
|
Provide a lower level __bin2hex function that does not allocate memory itself.
|
|
|
835ad824
|
2013-10-12T09:32:07
|
|
Convert the bitfury driver to use the hash_driver_work version of hash_work.
|
|
|
58008a37
|
2013-10-12T09:05:58
|
|
Add a hash_driver_work function to allow for drivers that wish to do their own work queueing and management.
|
|
|
3b2630cf
|
2013-10-12T08:45:49
|
|
Merge branch 'master' of github.com:ckolivas/cgminer
|
|
|
2ca8d859
|
2013-10-12T08:45:03
|
|
Convert all usb control transfers to asynchronous communication with our own timeout management as well.
|
|
|
159673cf
|
2013-10-11T06:26:58
|
|
Merge pull request #502 from kanoi/master
Klondike temp checking and read buffer size
|
|
|
595dcdfa
|
2013-10-04T14:56:33
|
|
Klondike - increase circular read buffer size
|
|
|
8e82da43
|
2013-10-04T10:57:07
|
|
Klondike - extra zero value and range checking in temp conversion
|
|
|
44f96c64
|
2013-10-10T07:56:41
|
|
klondike - display MHz also
|
|
|
b5379760
|
2013-10-11T23:06:40
|
|
Make pthread conditional timeouts handle all bulk usb transfer timeouts performing libusb_cancel_transfer, disabling timeouts within libusb itself.
|
|
|
cda797d1
|
2013-10-11T16:04:23
|
|
Avoid calling get_statline_before on exit to avoid trying to use it on drivers in an indeterminate state.
|
|
|
d437de84
|
2013-10-11T15:58:29
|
|
Avoid calling get_statline on exit.
|
|
|
03ad59f2
|
2013-10-11T15:37:55
|
|
Add a small amount to the usb timeout before cancelling to allow for a regular usb polling interval to pass.
|
|
|
08d8612f
|
2013-10-11T14:47:01
|
|
Do not attempt to clear a usb halt before sending the cancel message since all transfers should normally be cancelled before attempting to clear a halt condition, and only change the return message to a timeout if it's consistent with a cancellation.
|
|
|
6d18ce4c
|
2013-10-11T14:37:40
|
|
Retry up to USB_RETRY_MAX times to clear a halt condition before failing.
|
|
|
3a7a677f
|
2013-10-11T13:43:16
|
|
Show the error number as well as the description in erroring bulk transfers.
|
|
|
7fbe999f
|
2013-10-11T12:00:39
|
|
Drop logging level for failed to connect to stratum to verbose mode only since we hit it regularly.
|
|
|
a1391961
|
2013-10-11T11:59:35
|
|
Update NEWS.
|
|
|
b5dfacef
|
2013-10-11T10:15:06
|
|
We are always dependent on libusb handling events so use the blocking libusb_handle_events in the polling thread and use a bool to know if we should continue polling.
|
|
|
853aa28f
|
2013-10-11T08:49:59
|
|
Merge branch 'master' into async
Conflicts:
usbutils.c
|
|
|
b30ae4a2
|
2013-10-09T23:06:46
|
|
Use fractional hashrate return values in bitfury_scanhash to minimise the number of times we return 0 based on hashrate so far to further damp out displayed hashrate.
|
|
|
32bc7461
|
2013-10-09T22:17:56
|
|
Check for presence of driver name in DRIVER_COUNT_FOUND to prevent strcmp on a null pointer when a driver is not built in.
|
|
|
a715d27a
|
2013-10-08T20:26:57
|
|
Merge pull request #501 from kanoi/cmr
Cairnsmore1 support clock/identify commands via the API
|
|
|
d97c3a4b
|
2013-10-09T14:17:47
|
|
API-README update pgaset, pgaidentify and ascset docs
|
|
|
39ac69c1
|
2013-10-09T14:09:47
|
|
CMR allow sending flash and clock commands
|
|
|
2834eb8d
|
2013-10-08T23:35:22
|
|
Kill off threads that have failed using hash_sole_work instead of just disabling them.
|
|
|
bdbe9dd5
|
2013-10-08T23:26:23
|
|
Make the bf1 getinfo size a macro
|
|
|
e05bfddc
|
2013-10-08T23:18:57
|
|
Failing to add_cgpu in bitfury should be a terminal failure.
|
|
|
078f7f47
|
2013-10-08T23:15:15
|
|
Check return values when attempting to open a BF1 device and set the msg size as a macro.
|
|
|
a5ed1771
|
2013-10-08T22:23:51
|
|
Revert "Further smooth out avalon hashrate by limiting the maximum number of nonces counted each cycle to 1 per miner, saving them for later cycles."
This reverts commit 7be59992c121a61e56402cec21508e39401fad6c.
Of dubious utility since a full nonce range is not scanned before loading more work.
|
|
|
7be59992
|
2013-10-08T22:17:00
|
|
Further smooth out avalon hashrate by limiting the maximum number of nonces counted each cycle to 1 per miner, saving them for later cycles.
|
|
|
dc9c3783
|
2013-10-08T22:05:01
|
|
Display errors on failed usb read and write and consider sequential IO errors a permanent failure.
|
|
|
a22f5bfd
|
2013-10-07T19:50:38
|
|
Use libusb's own error name function instead of hand coding the error names.
|
|
|
936fca4d
|
2013-10-07T20:38:24
|
|
Limit ms_tdiff to 1 hour as a sanity check.
|
|
|
39ec9518
|
2013-10-08T21:33:29
|
|
Enable the usb buffer in avalon driver.
|
|
|
efd8c441
|
2013-10-07T20:35:42
|
|
Check for async transfer variants of error messages.
|
|
|
2806f49c
|
2013-10-07T20:34:13
|
|
Revert "Check for the async variants of the libusb errors and retry up to retry max when trying to clear a pipe error/stall."
This reverts commit c7d5256b3657bd6d4f2f04f7288cd757bcbf77a5.
|
|
|
d34279b2
|
2013-10-08T20:48:26
|
|
Remove unused variables.
|
|
|
f80c9003
|
2013-10-08T20:46:36
|
|
Merge branch 'master' into async
Conflicts:
api.c
cgminer.c
miner.h
usbutils.c
util.c
|
|
|
f8e28f25
|
2013-10-08T20:42:36
|
|
Try switching pools if for some reason we end up with only idle pools and have ended up current_pool set to an idle one.
|
|
|
3b527f14
|
2013-10-08T20:41:13
|
|
Check a pool is stable for >5 mins before switching back to it.
|
|
|
ef8efd6c
|
2013-10-08T00:01:26
|
|
Minimise the time between dropping the read devlock and grabbing the write devlock to avoid tons of logging spam in the interim.
|
|
|
97b5dd08
|
2013-10-07T20:38:24
|
|
Limit ms_tdiff to 1 hour as a sanity check.
|
|
|
6c757c6c
|
2013-10-07T20:35:42
|
|
Check for async transfer variants of error messages.
|
|
|
85e0b3a5
|
2013-10-07T20:03:53
|
|
Check for the async variants of the libusb errors and retry up to retry max when trying to clear a pipe error/stall.
|
|
|
a79768b7
|
2013-10-07T19:50:38
|
|
Use libusb's own error name function instead of hand coding the error names.
|
|
|
74775e4e
|
2013-10-07T18:40:09
|
|
Check for libusb transfer stall error to be consistent with async IO errors returned for a halt condition.
|
|
|
38ae7093
|
2013-10-07T17:56:57
|
|
Check for continuous IO errors on USB and consider the device inactive if more than retry max.
|
|
|
7318ecee
|
2013-10-07T23:34:01
|
|
Make the devlock a cglock in usbutils and only grab the write lock for fundamental changes allowing us to send and receive transfers concurrently without lock contention.
|
|
|
d5e9e084
|
2013-10-07T15:37:34
|
|
Prevent overflows in us_tdiff and ms_tdiff.
|
|
|
8037eb15
|
2013-10-07T12:00:49
|
|
Change second initialise message on bitfury verbose mode.
|
|
|
8c0ea19e
|
2013-10-05T09:31:36
|
|
Submitting an ntime offset nonce needs to be done on a copy of the work instead of the original so abstract out shared components as much as possible, minimising strdups in copy_work and make submit_work_async work take copied work, cleaning up code in the process.
|
|
|
b6a97de6
|
2013-10-04T21:39:42
|
|
Provide a way for drivers to submit work that it has internally rolled the ntime value by returning the amount it has ntime rolled to be added.
|