Streamline create_jobs_task

[eandersson]

I simplified the code by merging process_mining_job and generate_additional_work into a single function, as their separation was unnecessary - they performed the same task except for how extranonce_2 was incremented in generate_additional_work. I moved the extranonce_2 logic into the while loop for better clarity and flow, and removed the unnecessary yielding mechanism

In addition this also changes how extranonce_2 is rolled between notifies.

New version

0 -> 1 -> 2 -> Notify -> 0 -> 1 -> 2 -> Notify -> 0 -> 1 -> 2 etc

Old version

0 -> 1 -> 2 -> Notify -> 0 -> 3 -> 4 -> Notify -> 0 -> 5 -> 6 etc

[mutatrum]

Looking at it now. There's a difference in where extranonce_2 is reset back to 0, in the old code it was inside the while (GLOBAL_STATE->stratum_queue.count < 1 && GLOBAL_STATE->abandon_work == 0) loop, in the new code it's outside this loop.

I logged the extranonce_2, and with the old code there's a weird pattern which I don't think is what was supposed to happen:

I (40738) create_jobs_task: extranonce_2: 3500000000000000
I (41238) create_jobs_task: extranonce_2: 3600000000000000
I (41738) create_jobs_task: extranonce_2: 3700000000000000
I (42238) create_jobs_task: extranonce_2: 3800000000000000
I (42308) stratum_task: rx: "mining.notify"
I (42438) create_jobs_task: extranonce_2: 0000000000000000
I (43238) create_jobs_task: extranonce_2: 3900000000000000
I (43738) create_jobs_task: extranonce_2: 3a00000000000000
I (44238) create_jobs_task: extranonce_2: 3b00000000000000

With the new code, it actually resets after a mining notify:

I (30067) create_jobs_task: extranonce_2: 2100000000000000
I (30567) create_jobs_task: extranonce_2: 2200000000000000
I (31067) create_jobs_task: extranonce_2: 2300000000000000
I (31567) create_jobs_task: extranonce_2: 2400000000000000
I (31737) stratum_task: rx: "mining.notify"
I (32067) create_jobs_task: extranonce_2: 0000000000000000
I (32567) create_jobs_task: extranonce_2: 0100000000000000
I (33067) create_jobs_task: extranonce_2: 0200000000000000
I (33567) create_jobs_task: extranonce_2: 0300000000000000

Is resetting the extranonce_2 a better approach for covering the space, or is that dictated by other fields anyways? Maybe even simpler to have it just continuously increase, and never reset.

Other than this, it looks good. Not sure why there was such a big chunk of code duplicated. Nice cleanup.

[eandersson]

Looking at it now. There's a difference in where extranonce_2 is reset back to 0, in the old code it was inside the while (GLOBAL_STATE->stratum_queue.count < 1 && GLOBAL_STATE->abandon_work == 0) loop, in the new code it's outside this loop.

I logged the extranonce_2, and with the old code there's a weird pattern which I don't think is what was supposed to happen:

I (40738) create_jobs_task: extranonce_2: 3500000000000000
I (41238) create_jobs_task: extranonce_2: 3600000000000000
I (41738) create_jobs_task: extranonce_2: 3700000000000000
I (42238) create_jobs_task: extranonce_2: 3800000000000000
I (42308) stratum_task: rx: "mining.notify"
I (42438) create_jobs_task: extranonce_2: 0000000000000000
I (43238) create_jobs_task: extranonce_2: 3900000000000000
I (43738) create_jobs_task: extranonce_2: 3a00000000000000
I (44238) create_jobs_task: extranonce_2: 3b00000000000000

With the new code, it actually resets after a mining notify:

I (30067) create_jobs_task: extranonce_2: 2100000000000000
I (30567) create_jobs_task: extranonce_2: 2200000000000000
I (31067) create_jobs_task: extranonce_2: 2300000000000000
I (31567) create_jobs_task: extranonce_2: 2400000000000000
I (31737) stratum_task: rx: "mining.notify"
I (32067) create_jobs_task: extranonce_2: 0000000000000000
I (32567) create_jobs_task: extranonce_2: 0100000000000000
I (33067) create_jobs_task: extranonce_2: 0200000000000000
I (33567) create_jobs_task: extranonce_2: 0300000000000000

Is resetting the extranonce_2 a better approach for covering the space, or is that dictated by other fields anyways? Maybe even simpler to have it just continuously increase, and never reset.

Other than this, it looks good. Not sure why there was such a big chunk of code duplicated. Nice cleanup.

Good callout. I'll just move it back to static for now.

[mutatrum]

LGTM, can just use a little more cleanup.