On Sun, 17 May 2020 at 17:48, Philippe Mathieu-Daudé <f4...@amsat.org> wrote: > > Various places ignore the MemTxResult indicator of > transaction failed. Some cases might be justified > (DMA?) while other are probably bugs. To avoid > ignoring transaction errors, suggestion is to mark > functions returning MemTxResult with > warn_unused_result attribute.
Not necessarily a bad idea, but don't we have an awful lot of places that ignore the result that we'd need to fix first? thanks -- PMM