diff options
Diffstat (limited to 'results/classifier/deepseek-r1:14b/output/hypervisor/1908781')
| -rw-r--r-- | results/classifier/deepseek-r1:14b/output/hypervisor/1908781 | 10 |
1 files changed, 10 insertions, 0 deletions
diff --git a/results/classifier/deepseek-r1:14b/output/hypervisor/1908781 b/results/classifier/deepseek-r1:14b/output/hypervisor/1908781 new file mode 100644 index 00000000..bb5d71ce --- /dev/null +++ b/results/classifier/deepseek-r1:14b/output/hypervisor/1908781 @@ -0,0 +1,10 @@ + +x86-64 not faulting when CS.L = 1 and CS.D = 1 + +In a UEFI application I accidentally created a code segment descriptor where both the L and D bits were 1. This is supposed to generate a GP fault (e.g. see page 2942 of https://software.intel.com/sites/default/files/managed/39/c5/325462-sdm-vol-1-2abcd-3abcd.pdf). When running with KVM a fault did indeed occur, but when not specifying any acceleration, no fault occurred. + +Let me know if you need me to develop a minimum example to debug from. At the moment it's all part of a slightly more complicated bit of code. + +Version: 5.2.0 (compiled from source) +Command line options: -smp cores=4 -m 8192 (plus whatever uefi-run adds to plug in OVMF and my UEFI application). +Environment: Ubuntu 20.04 on Ryzen 3700X \ No newline at end of file |