diff options
Diffstat (limited to 'results/scraper/launchpad-without-comments/1900919')
| -rw-r--r-- | results/scraper/launchpad-without-comments/1900919 | 14 |
1 files changed, 14 insertions, 0 deletions
diff --git a/results/scraper/launchpad-without-comments/1900919 b/results/scraper/launchpad-without-comments/1900919 new file mode 100644 index 00000000..bbc48780 --- /dev/null +++ b/results/scraper/launchpad-without-comments/1900919 @@ -0,0 +1,14 @@ +PXB selected as root bus incorrectly + +release: 4c41341af76cfc85b5a6c0f87de4838672ab9f89 + +qdev_device_add() will search for the "closest" bus possible, and bail out early if that bus is a root bus. pxb devices are considered root buses and so if you either +1. Add a PCI device on the QEMU command line *after* a pxb device, or +2. Add an integrated PCI device (like a watchdog) + +#1: -device pxb-pcie,id=cxl.0,bus=pcie.0,bus_nr=52 -device ahci,id=sata0,addr=0x8 +#2: -watchdog i6300esb -device pxb-pcie,id=cxl.0,bus=pcie.0,bus_nr=52 + +The PXB will get selected as the bus (instead of the real root bus) and this will cause an assertion failure with the message like "qemu-system-x86_64: -device ahci,id=sata0,addr=0x8: PCI: Only PCI/PCIe bridges can be plugged into pxb-pcie" + +I think this is relatively solvable in the code base by determining if a bus is an expander, and skipping it if so. However, I wonder if it makes more sense to just allow expanders to have endpoint devices. \ No newline at end of file |