diff options
Diffstat (limited to 'results/classifier/accel-gemma3:12b/vmm/2882')
| -rw-r--r-- | results/classifier/accel-gemma3:12b/vmm/2882 | 91 |
1 files changed, 91 insertions, 0 deletions
diff --git a/results/classifier/accel-gemma3:12b/vmm/2882 b/results/classifier/accel-gemma3:12b/vmm/2882 new file mode 100644 index 00000000..79fd0569 --- /dev/null +++ b/results/classifier/accel-gemma3:12b/vmm/2882 @@ -0,0 +1,91 @@ + +Reading ACPI info via fw_cfg in SVSM causes Linux to panic +Description of problem: +We could use some help from a Qemu expert with an Qemu/ACPI/Linux related problem, +working on Coconut SVSM. + +**See https://github.com/coconut-svsm/svsm/issues/646** + +Coconut has code to read ACPI information via fw_cfg, and extract the number of guest CPUs from that. +That code has been used in the past, but since igvm became the default launch method for SVSM, it +was only used in corner-cases, while the information were obtained in some other way (igmv parameter). +Now a commit switches back to the fw_cfg+ACPI method. The information returned by it is correct, but +strangely Linux (guest) is spitting out ACPI related errors and crashes in some cases before reaching user-space. We do not have any clue how this can be related other than through Qemu behavior. +There is no direct way how SVSM can influence the ACPI related behavior of the Linux +guest kernel. + +The problem seems to be caused by simply reading the ACPI data. + +Reverting the bad commit and simply calling the original fw_cfg acpi function causes problems for Linux. +Steps to reproduce: +Boot SVSM bases CVM. SVSM and OVMF boot OK, then Linux prints these errors in some scenarios panics: +``` +[...] +[ 1.857709] ACPI: Added _OSI(Processor Aggregator Device) +[ 1.859436] ACPI: 1 ACPI AML tables successfully acquired and loaded +[ 1.860867] ACPI Error: AE_BAD_ADDRESS, Unable to initialize fixed events (20240827/evevent-53) +[ 1.862709] ACPI: Unable to start the ACPI Interpreter +[ 1.863708] ACPI Error: Could not remove SCI handler (20240827/evmisc-251) +[ 1.864942] ACPI Error: AE_BAD_PARAMETER, Thread 2176690624 could not acquire Mutex [ACPI_MT +X_Namespace] (0x1) (20240827/utmutex-252) +[ 1.866715] ACPI Error: AE_BAD_PARAMETER, Thread 2176690624 could not acquire Mutex [ACPI_MTX_Tables] (0x2) (20240827/utmutex-252) +[ 1.869722] ACPI Error: Mutex [ACPI_MTX_Tables] (0x2) is not acquired, cannot release (20240 +827/utmutex-289) +[ 1.870826] iommu: Default domain type: Translated +[ 1.871710] iommu: DMA domain TLB invalidation policy: lazy mod +[...] +[ 2.672635] io scheduler bfq registered +[ 2.675679] atomic64_test: passed for x86-64 platform with CX8 and with SSE +[ 2.677596] shpchp: Standard Hot Plug PCI Controller Driver version: 0.4 +[ 2.679264] ------------[ cut here ]------------ +[ 2.680284] refcount_t: addition on 0; use-after-free. +[ 2.681477] WARNING: CPU: 3 PID: 1 at lib/refcount.c:25 refcount_warn_saturate+0xe5/0x110 +[ 2.683261] Modules linked in: +[ 2.683929] CPU: 3 UID: 0 PID: 1 Comm: swapper/0 Not tainted 6.13.6-200.fc41.x86_64 #1 +[ 2.685608] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS edk2-stable202502-39-gb +483f751 02/02/2022 +[ 2.687729] RIP: 0010:refcount_warn_saturate+0xe5/0x110 +[ 2.688853] Code: e3 7f ff 0f 0b e9 fb 0a 8a 00 80 3d 15 9f 23 02 00 0f 85 5e ff ff ff 48 c7 c7 30 7b e7 8c c6 05 01 9f 23 02 01 e8 fb e2 7f ff <0f> 0b e9 d4 0a 8a 00 48 c7 c7 88 7b e7 8c + c6 05 e5 9e 23 02 01 e8 +[ 2.692768] RSP: 0018:ffffb2ed0001fd90 EFLAGS: 00010282 +[ 2.693894] RAX: 0000000000000000 RBX: ffff975b81060a80 RCX: ffffffff8d967448 +[ 2.695410] RDX: 0000000000000000 RSI: 0000000000000003 RDI: 0000000000000001 +[ 2.696923] RBP: ffffb2ed0001fe38 R08: 0000000000000000 R09: 0720072007200720 +[ 2.698439] R10: 0720072007200720 R11: 0720072007200720 R12: ffff975b81060a80 +[ 2.699955] R13: ffffb2ed0001fe78 R14: 00000000000000dc R15: 00000000000001df +[ 2.701461] FS: 0000000000000000(0000) GS:ffff975cf7d80000(0000) knlGS:0000000000000000 +[ 2.703179] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 +[ 2.704400] CR2: 00007f1c1658a3c8 CR3: 000800006082c000 CR4: 00000000003506f0 +[ 2.705910] Call Trace: +[ 2.706451] <TASK> +[ 2.706922] ? srso_return_thunk+0x5/0x5f +[ 2.707820] ? show_trace_log_lvl+0x255/0x2f0 +[ 2.708783] ? show_trace_log_lvl+0x255/0x2f0 +[ 2.709712] ? kobject_get+0x68/0x70 +[ 2.710492] ? refcount_warn_saturate+0xe5/0x110 +[ 2.711480] ? __warn.cold+0x93/0xfa +[ 2.712268] ? refcount_warn_saturate+0xe5/0x110 +[ 2.713262] ? report_bug+0xff/0x140 +[ 2.714036] ? handle_bug+0x58/0x90 +[ 2.714779] ? exc_invalid_op+0x17/0x70 +[ 2.715617] ? asm_exc_invalid_op+0x1a/0x20 +[ 2.716526] ? refcount_warn_saturate+0xe5/0x110 +[ 2.717507] kobject_get+0x68/0x70 +[ 2.718266] kobject_add_internal+0x32/0x250 +[ 2.719196] kobject_add+0x96/0xc0 +[ 2.719923] kobject_create_and_add+0xa3/0xc0 +[ 2.720851] bgrt_init+0x77/0xc0 +[ 2.721578] ? __pfx_bgrt_init+0x10/0x10 +[ 2.722418] do_one_initcall+0x5b/0x310 +[ 2.723272] do_initcalls+0x147/0x170 +[ 2.724086] ? __pfx_kernel_init+0x10/0x10 +[ 2.725174] kernel_init_freeable+0xfb/0x130 +[ 2.726114] kernel_init+0x1a/0x140 +[ 2.726883] ret_from_fork+0x34/0x50 +[ 2.727679] ? __pfx_kernel_init+0x10/0x10 +[ 2.728580] ret_from_fork_asm+0x1a/0x30 +[ 2.729429] </TASK> +[ 2.729926] ---[ end trace 0000000000000000 ]--- +``` +Additional information: + |