Core Duo T2500 laptop (ThinkPad T60) xen-3.0.2-3.FC5 domU: FC5 running 2.6.17-1.2174_FC5xenU dom0: FC5 running 2.6.17-1.2174_FC5xen0 Oracle Database 10g Express Universal Edition installed from RPM oracle-xe-univ-10.2.0.1-1.0 works ok in dom0, but when trying to run in domU, kernel reports Aug 24 19:31:46 oraXen kernel: Bad pte = 006780a0, process = oracle, vm_flags = 70, vaddr = b7f23000 Aug 24 19:31:46 oraXen kernel: <c04473bf> vm_normal_page+0x99/0xb5 <c0447b57> unmap_vmas+0x1f1/0x5be Aug 24 19:31:46 oraXen kernel: <c044311a> __pagevec_lru_add_active+0x119/0x12b <c044b2b7> unmap_region+0x8f/0xff Aug 24 19:31:46 oraXen kernel: <c044bd64> do_munmap+0x15b/0x1b3 <c044cd3b> do_mmap_pgoff+0x2b7/0x657 Aug 24 19:31:46 oraXen kernel: <c04090e8> sys_mmap2+0x7d/0xa6 <c0404ba7> syscall_call+0x7/0xb Aug 24 19:31:46 oraXen kernel: Bad pte = 006780a0, process = oracle, vm_flags = 70, vaddr = b7f24000 Aug 24 19:31:46 oraXen kernel: <c04473bf> vm_normal_page+0x99/0xb5 <c0447b57> unmap_vmas+0x1f1/0x5be Aug 24 19:31:46 oraXen kernel: <c044b2b7> unmap_region+0x8f/0xff <c044bd64> do_munmap+0x15b/0x1b3 Aug 24 19:31:46 oraXen kernel: <c044cd3b> do_mmap_pgoff+0x2b7/0x657 <c04090e8> sys_mmap2+0x7d/0xa6 Aug 24 19:31:46 oraXen kernel: <c0404ba7> syscall_call+0x7/0xb ... and goes on with host of "Bad pte" entries, until I stop oracle. The process does not die by itself.
I'm having the same issue in Oracle Standard Edition on FC5 2.6.17-1.2187_FC5xenU.
Is this using a PV or FV domU? Can you reproduce on current kernels?
I was PV domain. I'm runnig it fine on recent FC6 kernels.
change QA contact
FC5 is now end-of-life, and it seems that this bug is OK on FC6 or later. If this bug re-surfaces on a later version, please re-open. Chris Lalancette