On 5/22/20 6:26 AM, Richard Henderson wrote: > On 5/21/20 12:59 PM, Philippe Mathieu-Daudé wrote: >> +++ b/hw/semihosting/Kconfig >> @@ -1,3 +1,5 @@ >> >> +# default is 'n' >> config SEMIHOSTING >> - bool >> + bool >> + depends on TCG >> diff --git a/target/arm/Kconfig b/target/arm/Kconfig >> new file mode 100644 >> index 0000000000..035592dd86 >> --- /dev/null >> +++ b/target/arm/Kconfig >> @@ -0,0 +1,2 @@ >> +config SEMIHOSTING >> + default y if TCG > > Did you answer my question about replicating the TCG test? If you did, I > missed it...
Shortly before posting this series: https://www.mail-archive.com/qemu-devel@nongnu.org/msg705992.html There is indeed something wrong with this patch, but I'm not seeing it. - semihosting only makes sense with TCG accelerator This is the "depends on TCG" - some targets (if built with TCG enabled) require semihosting This would be a per-target "select SEMIHOSTING if TCG" Another problem is some targets only require semihosting in system-mode; and this approach select it for both system/user modes. Paolo... help!