Skip to contentSkip to conversation

IB/hfi1: clear VM_MAYWRITE on read-only STATUS mmap

PATCH New Abdifatah Suruur 2 participants 2 messages Nov 18

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 IB/hfi1: clear VM_MAYWRITE on read-only STATUS mmap

hfi1_file_mmap() rejects a writable mapping of the device status page, but leaves VM_MAYWRITE set in the VMA flags. Userspace can map the page read-only and then upgrade it to writable with mprotect(), bypassing the check and writing to a device-owned status page.

Clear VM_MAYWRITE on this path, as the RTAIL and SUBCTXT_EGRBUF cases in the same function already do.

---

drivers/infiniband/hw/hfi1/file_ops.c +1 -0
index 680364fda856f..779dbe32c69da 100644
@@ -492,6 +492,7 @@ static int hfi1_file_mmap(struct file *fp, struct vm_area_struct *vma)
 			ret = -EPERM;
 			goto done;
 		}
+		flags &= ~VM_MAYWRITE;
 		memaddr = kvirt_to_phys((void *)dd->status);
 		memlen = PAGE_SIZE;
 		break;
#1 lore raw
Thank you for your contribution! Sashiko AI review found 1 potential issue to consider: - [Low] IB/hfi1: missing Fixes trailer for STATUS mmap VM_MAYWRITE fix It also noticed 1 related pre-existing…

Thank you for your contribution! Sashiko AI review found 1 potential issue to consider:
- [Low] IB/hfi1: missing Fixes trailer for STATUS mmap VM_MAYWRITE fix It also noticed 1 related pre-existing issue this patch does not introduce:
- [Critical] IB/hfi1: missing VM_WRITE check in hfi1_file_mmap() [1] --

commit acb6e09931f843c6bef13cb9b188808881bec93a
Author: Abdifatah Suruur <[email protected]>

Quoted 7 lines

IB/hfi1: clear VM_MAYWRITE on read-only STATUS mmap

hfi1_file_mmap() rejects a writable mapping of the device status page, but leaves VM_MAYWRITE set in the VMA flags. Userspace can map the page read-only and then upgrade it to writable with mprotect(), bypassing the check and writing to a device-owned status page.

Quoted 5 lines

Clear VM_MAYWRITE on this path, as the RTAIL and SUBCTXT_EGRBUF cases in the same function already do.

[Severity: Low]
Since this fixes the read-only STATUS mmap check added in commit 12220267645c ("IB/hfi: Protect against writable mmap"), consider adding:

Fixes 12220267645c ("IB/hfi: Protect against writable mmap")

drivers/infiniband/hw/hfi1/file_ops.c static int hfi1_file_mmap(struct file *fp, struct vm_area_struct *vma) · line 498

Quoted 12 lines
drivers/infiniband/hw/hfi1/file_ops.c +1 -0
index 680364fda856f..779dbe32c69da 100644
@@ -492,6 +492,7 @@ static int hfi1_file_mmap(struct file *fp, struct vm_area_struct *vma)
 			ret = -EPERM;
 			goto done;
 		}
+		flags &= ~VM_MAYWRITE;
 		memaddr = kvirt_to_phys((void *)dd->status);
 		memlen = PAGE_SIZE;
 		break;

[Severity: Critical]
This problem wasn't introduced by this patch, but can userspace map the SUBCTXT_EGRBUF region as writable [1] by passing PROT_WRITE to mmap()?

drivers/infiniband/hw/hfi1/file_ops.c:hfi1_file_mmap() {
	...
	case SUBCTXT_EGRBUF:
		memvirt = uctxt->subctxt_rcvegrbuf;
		memlen = uctxt->egrbufs.size * uctxt->subctxt_cnt;
		flags &= ~VM_MAYWRITE;
		is_vmalloc = 1;
		break;
	...
}

The SUBCTXT_EGRBUF case clears VM_MAYWRITE from flags, but unlike RCV_EGRBUF, STATUS, and RTAIL, it never checks if VM_WRITE is set in flags and does not reject the mapping with -EPERM. When userspace maps SUBCTXT_EGRBUF with PROT_WRITE, vm_flags_reset() leaves VM_WRITE set on the VMA and remap_vmalloc_range() maps uctxt->subctxt_rcvegrbuf with write permissions, allowing userspace to modify the read-only eager receive buffers.

[1] https://sashiko.dev/#/bug/linux-011f18cc-47f7-48a3-9cad-8eb2f835cdbe

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1