The err_restore_domain flow was accidently inserted into the success path
in commit
1000dccd5d13 ("iommu: Allow IOMMU_RESV_DIRECT to work on
ARM"). It should only happen if iommu_create_device_direct_mappings()
fails. This caused the domains the be wrongly changed and freed whenever
the sysfs is used, resulting in an oops:
BUG: kernel NULL pointer dereference, address:
0000000000000000
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 0 P4D 0
Oops: 0000 [#1] PREEMPT SMP NOPTI
CPU: 1 PID: 3417 Comm: avocado Not tainted 6.4.0-rc4-next-
20230602 #3
Hardware name: Dell Inc. PowerEdge R6515/07PXPY, BIOS 2.3.6 07/06/2021
RIP: 0010:__iommu_attach_device+0xc/0xa0
Code: c0 c3 cc cc cc cc 48 89 f0 c3 cc cc cc cc 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 0f 1f 44 00 00 41 54 55 48 8b 47 08 <48> 8b 00 48 85 c0 74 74 48 89 f5 e8 64 12 49 00 41 89 c4 85 c0 74
RSP: 0018:
ffffabae0220bd48 EFLAGS:
00010246
RAX:
0000000000000000 RBX:
ffff9ac04f70e410 RCX:
0000000000000001
RDX:
ffff9ac044db20c0 RSI:
ffff9ac044fa50d0 RDI:
ffff9ac04f70e410
RBP:
ffff9ac044fa50d0 R08:
1000000100209001 R09:
00000000000002dc
R10:
0000000000000000 R11:
0000000000000000 R12:
ffff9ac043d54700
R13:
ffff9ac043d54700 R14:
0000000000000001 R15:
0000000000000001
FS:
00007f02e30ae000(0000) GS:
ffff9afeb2440000(0000) knlGS:
0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0:
0000000080050033
CR2:
0000000000000000 CR3:
000000012afca006 CR4:
0000000000770ee0
PKRU:
55555554
Call Trace:
<TASK>
? __die+0x24/0x70
? page_fault_oops+0x82/0x150
? __iommu_queue_command_sync+0x80/0xc0
? exc_page_fault+0x69/0x150
? asm_exc_page_fault+0x26/0x30
? __iommu_attach_device+0xc/0xa0
? __iommu_attach_device+0x1c/0xa0
__iommu_device_set_domain+0x42/0x80
__iommu_group_set_domain_internal+0x5d/0x160
iommu_setup_default_domain+0x318/0x400
iommu_group_store_type+0xb1/0x200
kernfs_fop_write_iter+0x12f/0x1c0
vfs_write+0x2a2/0x3b0
ksys_write+0x63/0xe0
do_syscall_64+0x3f/0x90
entry_SYSCALL_64_after_hwframe+0x6e/0xd8
RIP: 0033:0x7f02e2f14a6f
Reorganize the error flow so that the success branch and error branches
are clearer.
Fixes:
1000dccd5d13 ("iommu: Allow IOMMU_RESV_DIRECT to work on ARM")
Reported-by: Dheeraj Kumar Srivastava <dheerajkumar.srivastava@amd.com>
Tested-by: Vasant Hegde <vasant.hegde@amd.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Link: https://lore.kernel.org/r/0-v1-5bd8cc969d9e+1f1-iommu_set_def_fix_jgg@nvidia.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
ret = __iommu_group_set_domain_internal(
group, dom, IOMMU_SET_DOMAIN_MUST_SUCCEED);
if (WARN_ON(ret))
- goto out_free;
+ goto out_free_old;
} else {
ret = __iommu_group_set_domain(group, dom);
- if (ret) {
- iommu_domain_free(dom);
- group->default_domain = old_dom;
- return ret;
- }
+ if (ret)
+ goto err_restore_def_domain;
}
/*
for_each_group_device(group, gdev) {
ret = iommu_create_device_direct_mappings(dom, gdev->dev);
if (ret)
- goto err_restore;
+ goto err_restore_domain;
}
}
-err_restore:
- if (old_dom) {
+out_free_old:
+ if (old_dom)
+ iommu_domain_free(old_dom);
+ return ret;
+
+err_restore_domain:
+ if (old_dom)
__iommu_group_set_domain_internal(
group, old_dom, IOMMU_SET_DOMAIN_MUST_SUCCEED);
+err_restore_def_domain:
+ if (old_dom) {
iommu_domain_free(dom);
- old_dom = NULL;
+ group->default_domain = old_dom;
}
-out_free:
- if (old_dom)
- iommu_domain_free(old_dom);
return ret;
}