On 7/6/21 10:24 AM, Thomas Huth wrote: > On 16/04/2021 14.52, Thomas Huth wrote: >> QEMU currently crashes when the user tries to do something like: >> >> qemu-system-x86_64 -M x-remote -device piix3-ide > > It's now several months later already, and this crash has still not been > fixed yet. The next softfreeze is around the corner and the > "device-crash-test" still stumbles accross this problem. > If the other solutions are not mergable yet (what's the status here?),
See this big thread about ISA vs PCI IDE modelling / design: https://www.mail-archive.com/qemu-devel@nongnu.org/msg809678.html TL;DR: short term we are screwed. long term, not worth it. Stefan, IIRC the multi-process conclusion was we have to reject PCI devices briding another (non-PCI) bus, such ISA / I2C / USB / SD / ... because QEMU register the bus type globally and the command line machinery resolves it to plug user-creatable devices, so we can not share such buses. Is that correct? > could we please include my patch for QEMU v6.1 instead, so that we get > this silenced in the device-crash-test script? Yes please. Regards, Phil. >> This happens because the "isabus" variable is not initialized with >> the x-remote machine yet. Add a proper check for this condition >> and propagate the error to the caller, so we can fail there gracefully. >> >> Signed-off-by: Thomas Huth <th...@redhat.com> >> --- >> hw/ide/ioport.c | 16 ++++++++++------ >> hw/ide/piix.c | 22 +++++++++++++++++----- >> hw/isa/isa-bus.c | 14 ++++++++++---- >> include/hw/ide/internal.h | 2 +- >> include/hw/isa/isa.h | 13 ++++++++----- >> 5 files changed, 46 insertions(+), 21 deletions(-)
