Integer overflow in numSamplesSqr calculation in ecmProcessSet

[FredM67]

Description

In src/emon_CM.c at line 789, the calculation of numSamplesSqr can overflow:

const int32_t numSamples    = accumProcessing->numSamples;
const int64_t numSamplesSqr = numSamples * numSamples;

The multiplication numSamples * numSamples is performed as int32_t * int32_t, yielding an int32_t result before being assigned to the int64_t variable. If numSamples exceeds ~46340, the multiplication overflows.

Expected behavior

The multiplication should be performed in 64-bit to prevent overflow:

const int64_t numSamplesSqr = (int64_t)numSamples * numSamples;

Or using unsigned types:

const uint32_t numSamples    = accumProcessing->numSamples;
const uint64_t numSamplesSqr = (uint64_t)numSamples * numSamples;

Fix

This issue will be fixed along with PR #40.

[awjlogan]

Makes sense - I'd expected the promotion to be done based on the l-value type, but clearly not. I only tested out to a 10 minute report period, so unlikely to be hit in the wild but let's do it :)

[FredM67]

I'll have maybe a surprise later on :D

[awjlogan]

Haha, I don't need surprises right now 😂 😢

[FredM67]

Even good surprises ?

[awjlogan]

Who knew such things existed 😛 Alright, I'll await with anticipation!

[FredM67]

Who knew such things existed 😛 Alright, I'll await with anticipation!

See PR#47