I was able to reproduce it at my end. This problem was introduced in a recent refactor. The problem was that the code for configuring the VMM-side of cross vm connections (including those used by string_reverse) was not being linked into the VMM, so the required hypercall handlers were not registered.

From: Devel <devel-bounces@sel4.systems> on behalf of Michal Podhradsky <mpodhradsky@galois.com>
Sent: Saturday, 20 May 2017 9:08 AM
To: devel@sel4.systems
Subject: [seL4] Fwd: k_vmm_manager: failed on sel4 hypervisor connection

Hello all,

I am running camkes-vm on x86 architecture, following the example on wiki https://wiki.sel4.systems/CAmkESVM#Cross_VM_Connectors with the exception that I want to invoke the stringReverse component.

I am running the optiplex9020_defconfig example with only one VM, and when I start booting the guest linux system, I see the following error:

[ 1.282326] dataport: loading out-of-tree module taints kernel.

[ 1.290615] dataport: module license 'unspecified' taints kernel.

[ 1.306854] Disabling lock debugging due to kernel taint

[ 1.319664] dataport initialized with major number 247

vmm_vmcall_handler:55 | Failed to find handler for token:2

[ 1.325411] incorrect event context magic number (expected 42, got -677408512)insmod: can't insert '/lib/modules/4.8.16/kernel/drivers/vmm/consumes_event.ko': Operation not ermitted

[ 1.337441] emits_event initialized with major number 245

[ 1.356391] k_vmm_manager: failed on sel4 hypervisor connection |-1|

Then when I log into linux and call `string_reverse` program, I get the following error:

# string_reverse

[ 8.360164] BUG: unable to handle kernel paging request at d881f2c0

[ 8.360164] IP: [<d00a69f0>] chrdev_open+0xab/0x100

[ 8.360164] *pde = 17867067 *pte = 00000000

[ 8.360164] Oops: 0000 [#1]

[ 8.360164] Modules linked in: emits_event(PO) dataport(PO)

[ 8.360164] CPU: 0 PID: 691 Comm: string_reverse Tainted: P O 4.8.16 #1

[ 8.360164] task: d79c9740 task.stack: d6ae0000

[ 8.360164] EIP: 0060:[<d00a69f0>] EFLAGS: 00010286 CPU: 0

[ 8.360164] EIP is at chrdev_open+0xab/0x100

[ 8.360164] EAX: d881f2c0 EBX: 00000000 ECX: d79b0d00 EDX: 00000000

[ 8.360164] ESI: d79b0d00 EDI: d6add808 EBP: d6ae1e10 ESP: d6ae1df8

[ 8.360164] DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 0068

[ 8.360164] CR0: 80050033 CR2: d881f2c0 CR3: 16ad9000 CR4: 00102690

[ 8.360164] DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000

[ 8.360164] DR6: fffe0ff0 DR7: 00000400

[ 8.360164] Stack:

[ 8.360164] d7ac9be0 d7ac9be0 00000001 d7ac9be0 d6add808 00000000 d6ae1e30 d00a2370

[ 8.360164] d6ae1e28 d7ac9be8 d00a6945 d7ac9be0 d69db010 d7462d80 d6ae1e4c d00a2cf8

[ 8.360164] d79c2b80 d79c2b80 00000000 d6ae1ebc d7462d80 d6ae1eb0 d00ae159 d6ae1e64

[ 8.360164] Call Trace:

[ 8.360164] [<d00a2370>] do_dentry_open+0x197/0x216

[ 8.360164] [<d00a6945>] ? cdev_put+0x1a/0x1a

[ 8.360164] [<d00a2cf8>] vfs_open+0x45/0x4e

[ 8.360164] [<d00ae159>] path_openat+0x822/0x970

[ 8.360164] [<d007a468>] ? __alloc_pages_nodemask+0x11c/0x770

[ 8.360164] [<d00ae2d8>] do_filp_open+0x31/0x77

[ 8.360164] [<d00b6293>] ? __alloc_fd+0xbb/0xfe

[ 8.360164] [<d00a2e56>] do_sys_open+0x54/0xc7

[ 8.360164] [<d00a2ee1>] SyS_open+0x18/0x1a

[ 8.360164] [<d0000ee1>] do_fast_syscall_32+0x7a/0xb8

[ 8.360164] [<d047831f>] sysenter_past_esp+0x40/0x6a

[ 8.360164] Code: 11 89 f0 e8 e1 fa ff ff 83 f8 01 19 db 31 d2 83 e3 fa 89 d0 e8 4d ff ff ff 85 db 75 59 8b 46 28 85 c0 75 0c

[ 8.360164] EIP: [<d00a69f0>] chrdev_open+0xab/0x100 SS:ESP 0068:d6ae1df8

[ 8.360164] CR2: 00000000d881f2c0

[ 8.360164] ---[ end trace 46873a57a343e435 ]---

[ 8.376285] string_reverse (691) used greatest stack depth: 6496 bytes left

Killed

which seems like the program is trying to access memory that it is not supposed to.

If I follow the example from wiki (here https://wiki.sel4.systems/CAmkESVM#Cross_VM_Connectors ) I get the same error, but triggered by `print_client` program.

Have you seen a problem like this before? How do I correctly load the kernel module?

Regards

Michal