dangernoodle-io / TaipanMiner / 25282460041

90%

push

github

fix(stratum): collapse two 4 KB line buffers into one (TA-298) (#311)

stratum_io kept a 4 KB file-scope `s_linebuf` recv accumulator AND a
parallel 4 KB `char line[4096]` on the stratum_task stack. Each socket
read accumulated into the static buffer, then a `memcpy` published the
extracted line into the stack buffer for `process_message`. Two copies of
the same data, costing 4 KB BSS + 4 KB stack permanently.

Refactor stratum_readline() to extract the line in-place: write the null
terminator into s_linebuf at the '\\n' position, return a pointer to the
buffer, and defer the post-line memmove to the next readline call (via
s_pending_consume). Caller now passes `const char **` instead of a
destination buffer.

Net win: drop the 4 KB stack array. BSS unchanged (s_linebuf must still
hold the largest mining.notify message, which can run several KB on pools
with deep merkle branches). A follow-up can shrink the 8 KB stratum task
stack now that we've freed 4 KB of it.

Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>

707 of 852 branches covered (82.98%)

Branch coverage included in aggregate %.

1607 of 1654 relevant lines covered (97.16%)

249583.16 hits per line