CVE-2026-90008: scsi: megaraid_sas: Limit NVMe request size to the PRP chain frame
In the Linux kernel, the following vulnerability has been resolved:
scsi: megaraidsas: Limit NVMe request size to the PRP chain frame
megasasmakeprpnvme() builds a command's PRP list in cmd->sgframe, a DMA pool buffer of instance->maxchainframesz bytes, spending one entry per NVMe page of the transfer plus one per page of the buffer for the chain pointer. The loop runs until the transfer is described and never checks the buffer bound.
maxhwsectors comes straight from the MDTS the firmware reports for the drive. On drives with a large MDTS the only thing keeping the list inside the buffer was the block layer default of 1280 KiB, which needs 320 entries, which fit into a 4 KiB frame as that holds 512. But since commit 9b8b84879d4a ("block: Increase BLKDEFMAXSECTORSCAP") that default is 4 MiB, and such a transfer needs 1025 entries, so the list runs a full page past the end of the frame:
sd 1:0:1:0: [sdb] tag#630 page boundary ptrsgl: 0x00000000ba62d13f BUG: unable to handle page fault for address: ff663bcb81e7c000 #PF: supervisor write access in kernel mode #PF: errorcode(0x0002) - not-present page RIP: 0010:megasasbuildandissuecmdfusion+0xeaa/0x1870 [megaraidsas]
If the page after the frame happens to be mapped, the overrun does not fault but silently corrupts the neighbouring pool entry, which is another in-flight command's PRP list.
Cap maxhwsectors at what the chain frame can describe, less one page for transfers that do not start on a page boundary and so need one entry more. This is the megaraidsas counterpart of commit 04631f55afc5 ("scsi: mpt3sas: Limit NVMe request size to 2 MiB"), but derives the limit from maxchainframesz rather than hardcoding it.
Affected Software
Event History
Frequently Asked Questions
Which systems are exposed to this issue?
Systems using the Linux megaraid_sas driver for NVMe I/O may be affected when the drive firmware reports a large MDTS and requests can exceed the PRP-chain frame capacity. The described failure occurs with a 4 MiB transfer requiring 1025 entries when a 4 KiB frame holds only 512 entries.
Is the default block-layer request limit relevant?
Yes. The earlier 1280 KiB default required 320 entries and fit within the described 4 KiB frame, while the increased 4 MiB default can overrun that frame by a full page under the stated conditions.
How might an affected system present?
The reported symptom is a kernel-mode supervisor write page fault during I/O, with megasas_build_and_issue_cmd_fusion in the faulting stack. Logs may also show a page-boundary pointer message for an sd device before the fault.