platform: stm32: fix SPI DMA timeout underflow

[morleman1]

This fixes an issue where an integer underflow in the timeout loop masks DMA failures and creates a race condition that reports false timeouts.

Resolves #3115

Pull Request Description

Problem

The loop while (timeout--) in stm32_spi_transfer_dma() causes two issues:

1. Integer Underflow: If DMA stalls, timeout decrements past 0 to 0xFFFFFFFF. The final if (timeout == 0) check fails, and the function incorrectly returns 0 (Success).
2. Race Condition: If DMA completes during the very last iteration, the loop breaks but timeout is already 0, causing a false -ETIME return.

Solution

• Changed loop condition to while (timeout > 0) to prevent underflow.
• Verified completion using the actual hardware flag !sdesc->stm32_spi_dma_done instead of the counter.

Testing

Verified via hardware debugging on STM32 with an ADIN1110 by simulating a DMA stall. The driver now correctly catches the stall and returns -ETIME.

PR Type

• Bug fix (change that fixes an issue)
• New feature (change that adds new functionality)
• Breaking change (has dependencies in other repos or will cause CI to fail)

PR Checklist

• I have followed the Coding style guidelines
• I have complied with the Submission Checklist
• I have performed a self-review of the changes
• I have commented my code, at least hard-to-understand parts
• I have build all projects affected by the changes in this PR
• I have tested in hardware affected projects, at the relevant boards
• I have signed off all commits from this PR
• I have updated the documentation (wiki pages, ReadMe etc), if applies

[CLAassistant]

All committers have signed the CLA.

[buha]

Thanks for catching it and submitting the fix. While you are at it, and because they seem identical, could you fix the same issue in xspi implementation ? You can put the xspi change in the same commit as they are tightly related.
https://github.com/analogdevicesinc/no-OS/blob/main/drivers/platform/stm32/stm32_xspi.c#L696

[buha]

remove this comment as well

[morleman1]

i reverted to the original comment /* need some cleanup here? */, as a cleanup still should be implemented similar to the xspi implementation. here it is more complex, because of seperate handles for rx and txdma_ch

[buha]

You are right, just squash the commits for now into a single commit and leave the original question there. We can deal with that separately in another PR and merge this one.