On 19/03/20 6:14 pm, Rasmus Villemoes wrote:
Hmm, Watchdog seems to be set to trigger after ~2s of inactivity. Isn't that too small? WATCHDOG_RESET() resets only once per second, so 2 second timeout is too close IMO.
Typical watchdog timeouts are usually in tens of seconds
Nope, not with this external gpio-triggered one. The data sheet says
Watchdog timeout period 1.12/1.60/2.24 (min/typ/max)
so ~2 seconds sounds about right - and I have to account for other instances of the board that may be a lot closer to the minimum. The timeout is fixed, nothing in software can affect it. So you see why I cannot afford to let spi flash operations take several seconds to complete without a WATCHDOG_RESET().
And yes, I'm very well aware of the rate-limiting imposed by the wdt-provided watchdog_reset() function - that's mostly a solved problem: https://patchwork.ozlabs.org/project/uboot/list/?series=164254
Understood.
For the read side, it seems that just replacing the UINT_MAX in the two copies of spi_nor_read_data with some smaller constant should suffice. But I don't know if I should make that smaller constant a CONFIG_* option or just pick something like 256K. Thoughts?
Breaking reads into smaller units unconditionally will cause performance regressions. But I would like to avoid adding new CONFIG option as well.
How about resetting the watchdog in the SPI driver's read implementation? Or setting max_read_size (in struct spi_slave) to smaller value in the SPI controller driver to force fragmentation of reads?