Hi Igor, One additional observation: QEMU's x86 ISA serial implementation explicitly defines _STA = 0x0f in serial_isa_build_aml(), rather than relying on the ACPI default:
https://github.com/qemu/qemu/blob/8967b2fe1b2aaae3c29e207d85fab901d580a2dc/hw/char/serial-isa.c#L101 I think following the same approach for LoongArch UART would be reasonable for consistency. Thanks, pillow On Thu, Oct 22, 2026 at 5:01 PM way sorry <[email protected]> wrote: > Hi Igor, > > The _STA method was added in response to Bibo's review comment on the > original patch, where > he explicitly suggested adding _STA. > > I understand your concern that omitting _STA already implies 0x0f. > However, the two review > comments seem to suggest different approaches. > > I'm happy to split the changes into separate patches as required. For > _STA, though, I'd > appreciate it if we could first agree on whether it should be added or > omitted, so I have a clear > direction for the next version. > > Thanks, > pillow > > On Thu, Oct 22, 2026 at 4:18 PM Igor Mammedov <[email protected]> wrote: > >> On Thu, 22 Oct 2026 15:22:36 +0800 >> pillow <[email protected]> wrote: >> >> ... >> > @@ -325,7 +324,7 @@ static void build_uart_device_aml(Aml *table, int >> index) >> ... >> > + aml_append(dev, aml_name_decl("_STA", aml_int(0x0f))); >> >> also is it really necessary? Doesn't missing _STA imply 0xF already? >> >> > crs = aml_resource_template(); >> > aml_append(crs, >> > aml_qword_memory(AML_POS_DECODE, AML_MIN_FIXED, AML_MAX_FIXED, >> >>
