OESA-2026-1831

The Linux Kernel, the operating system core itself.

Security Fix(es):

A vulnerability exists in the RDMA/umad component of the Linux kernel when processing user-space data. The ibumadwrite function calculates datalen based on user-controlled count and MAD header sizes. When a mismatched user MAD header size and RMPP header length are provided, datalen can become negative and be passed to the ibcreatesendmad() function. This can cause subsequent padding calculations to exceed the segment size, ultimately triggering an out-of-bounds memory write (memset) in the allocsendrmpplist() function. An attacker could potentially exploit this vulnerability to cause kernel memory corruption, affecting system stability and confidentiality.(CVE-2026-23243)

In the Linux kernel, the following vulnerability has been resolved:

x86/efi: defer freeing of boot services memory

efifreebootservices() frees memory occupied by EFIBOOTSERVICESCODE and EFIBOOTSERVICESDATA using memblockfree_late().

There are two issue with that: memblockfreelate() should be used for memory allocated with memblockalloc() while the memory reserved with memblockreserve() should be freed with freereservedarea().

More acutely, with CONFIGDEFERREDSTRUCTPAGEINIT=y efifreeboot_services() is called before deferred initialization of the memory map is complete.

Benjamin Herrenschmidt reports that this causes a leak of ~140MB of RAM on EC2 t3a.nano instances which only have 512MB or RAM.

If the freed memory resides in the areas that memory map for them is still uninitialized, they won't be actually freed because memblockfreelate() calls memblockfreepages() and the latter skips uninitialized pages.

Using freereservedarea() at this point is also problematic because _freepage() accesses the buddy of the freed page and that again might end up in uninitialized part of the memory map.

Delaying the entire efifreeboot_services() could be problematic because in addition to freeing boot services memory it updates efi.memmap without any synchronization and that's undesirable late in boot when there is concurrency.

More robust approach is to only defer freeing of the EFI boot services memory.

Split efifreebootservices() in two. First efiunmapbootservices() collects ranges that should be freed into an array then efifreeboot_services() later frees them after deferred init is complete.(CVE-2026-23352)

In the Linux kernel, a vulnerability has been resolved in Squashfs component: check metadata block offset is within range. Syzkaller reports a "general protection fault in squashfscopydata". This is ultimately caused by a corrupted index look-up table, which produces a negative metadata block offset. This is subsequently passed to squashfscopydata (via squashfsreadmetadata) where the negative offset causes an out of bounds access. The fix is to check that the offset is within range in squashfsreadmetadata. This will trap this and other cases.(CVE-2026-23388)