[BUG] topology1: FW pcm_params fail with large period size >44739

[kv2019i]

Describe the bug
Continuation of
#7476

See log in #7476 (comment)

[kv2019i]

So this error

[      119612.963997] (          14.843749) c1 buffer                   src/audio/buffer.c:55   ERROR buffer_alloc(): could not alloc size = 4294963712 bytes of type = 32
[      119633.536913] (          20.572916) c1 dma-trace                 src/trace/trace.c:141  INFO Suppressed 22 similar messages:    Number of Blocks: total % used %d free %d
[      119646.766079] (          13.229166) c1 host         1.0           src/audio/host.c:845  ERROR host_params(): failed to alloc dma buffer
[      119669.213995] (          22.447916) c1 pipe         1.7   ....../pipeline-params.c:268  ERROR pipeline_params(): ret = -12, host->comp.id = 0

It seems we have two problems here:

• component_set_nearest_period_frames() in SOF FW component.h overflows as it does ceil_divide() in signed 32bit arithmetic (i.e. same bug we had on toploogy side, is on FW side as well)
• for some reason, while the tplg conf file has 50000 period size, kernel log in [BUG] toplogy1 : error: tuple SOF_TKN_BUF_SIZE: Numerical result out of range when pipeline period size increases beyond 40ms #7476 has "pipeline PIPELINE.29.SSP1.IN: period 80000 pri 0 mips 5000 core 1 frames 0 dynamic 0"

@Vamshigopal where the conf file and dmesg a matching pair? Is the period size supposed to be 50ms or 80ms?

I assume the logs files were just different, so that leaves us with the problem with FW calculating buffer size incorrectly if SOF_TKN_SCHED_PERIOD > 44739.

[Vamshigopal]

@kv2019i i took the conf file and dmesg at different instances, so im not sure, let me share the logs and conf file with 50ms period size.
i do agree with your assumption, as i see same error for 50ms, 80ms.. i.e SOF_TKN_SCHED_PERIOD > 44739

[kv2019i]

@Vamshigopal Bugfix candidate at #7655
I didn't realize we had similar signed 32bit arithmetic in FW to calculate the frame size. :( So this obviously fails in the same conditions as the m4 calculations failed.