diff options
Diffstat (limited to 'results/classifier/105/device/812398')
| -rw-r--r-- | results/classifier/105/device/812398 | 38 |
1 files changed, 38 insertions, 0 deletions
diff --git a/results/classifier/105/device/812398 b/results/classifier/105/device/812398 new file mode 100644 index 000000000..12beac7f9 --- /dev/null +++ b/results/classifier/105/device/812398 @@ -0,0 +1,38 @@ +device: 0.792 +instruction: 0.606 +vnc: 0.601 +graphic: 0.525 +mistranslation: 0.519 +socket: 0.510 +other: 0.504 +network: 0.458 +semantic: 0.448 +boot: 0.382 +assembly: 0.160 +KVM: 0.125 + +powerpc 7450 MMU initialization broken + +The 7540 family of PPCs' MMU can update TLBs using hardware search (like a 604 or 7400) but also using a software algorithm. The mechanism used is defined by HID0[STEN]. + +By default (CPU reset) HID0 is set to 0x80000000 (BTW; another small bug, qemu doesn't set the hardwired MSB), hence +the software-table lookup feature is *disabled*. However, the default (and immutable) 'mmu_model' for this CPU family is POWERC_MMU_SOFT_74XX which choses the soft TLB replacement scheme. + +To fix this: + +1) the initial mmu_model for the 7450 family (includes 7441, 7445, 7451, 7455, 7457, 7447, 7448) should be: POWERPC_MMU_32B +2) when HID0[STEN] is written then the mmu_model should be changed accordingly (I'm not familiar enough with the qemu internal state to judge if any cached state would have to be updated). + +Looking through old bug tickets... is this still an issue with the latest version of QEMU? Or could we close this ticket nowadays? + + +From looking at the source code of 5.1.0-rc3 (target/ppc/translate_init.inc.c) it seems that this is still an issue. + + +This is an automated cleanup. This bug report has been moved to QEMU's +new bug tracker on gitlab.com and thus gets marked as 'expired' now. +Please continue with the discussion here: + + https://gitlab.com/qemu-project/qemu/-/issues/86 + + |