{"id":"AZL-73710","summary":"CVE-2025-39691 affecting package kernel for versions less than 5.15.200.1-1","details":"In the Linux kernel, the following vulnerability has been resolved:\n\nfs/buffer: fix use-after-free when call bh_read() helper\n\nThere's issue as follows:\nBUG: KASAN: stack-out-of-bounds in end_buffer_read_sync+0xe3/0x110\nRead of size 8 at addr ffffc9000168f7f8 by task swapper/3/0\nCPU: 3 UID: 0 PID: 0 Comm: swapper/3 Not tainted 6.16.0-862.14.0.6.x86_64\nHardware name: QEMU Standard PC (i440FX + PIIX, 1996)\nCall Trace:\n \u003cIRQ\u003e\n dump_stack_lvl+0x55/0x70\n print_address_description.constprop.0+0x2c/0x390\n print_report+0xb4/0x270\n kasan_report+0xb8/0xf0\n end_buffer_read_sync+0xe3/0x110\n end_bio_bh_io_sync+0x56/0x80\n blk_update_request+0x30a/0x720\n scsi_end_request+0x51/0x2b0\n scsi_io_completion+0xe3/0x480\n ? scsi_device_unbusy+0x11e/0x160\n blk_complete_reqs+0x7b/0x90\n handle_softirqs+0xef/0x370\n irq_exit_rcu+0xa5/0xd0\n sysvec_apic_timer_interrupt+0x6e/0x90\n \u003c/IRQ\u003e\n\n Above issue happens when do ntfs3 filesystem mount, issue may happens\n as follows:\n           mount                            IRQ\nntfs_fill_super\n  read_cache_page\n    do_read_cache_folio\n      filemap_read_folio\n        mpage_read_folio\n\t do_mpage_readpage\n\t  ntfs_get_block_vbo\n\t   bh_read\n\t     submit_bh\n\t     wait_on_buffer(bh);\n\t                            blk_complete_reqs\n\t\t\t\t     scsi_io_completion\n\t\t\t\t      scsi_end_request\n\t\t\t\t       blk_update_request\n\t\t\t\t        end_bio_bh_io_sync\n\t\t\t\t\t end_buffer_read_sync\n\t\t\t\t\t  __end_buffer_read_notouch\n\t\t\t\t\t   unlock_buffer\n\n            wait_on_buffer(bh);--\u003e return will return to caller\n\n\t\t\t\t\t  put_bh\n\t\t\t\t\t    --\u003e trigger stack-out-of-bounds\nIn the mpage_read_folio() function, the stack variable 'map_bh' is\npassed to ntfs_get_block_vbo(). Once unlock_buffer() unlocks and\nwait_on_buffer() returns to continue processing, the stack variable\nis likely to be reclaimed. Consequently, during the end_buffer_read_sync()\nprocess, calling put_bh() may result in stack overrun.\n\nIf the bh is not allocated on the stack, it belongs to a folio.  Freeing\na buffer head which belongs to a folio is done by drop_buffers() which\nwill fail to free buffers which are still locked.  So it is safe to call\nput_bh() before __end_buffer_read_notouch().","modified":"2026-04-01T05:22:19.757012Z","published":"2025-09-05T18:15:45Z","upstream":["CVE-2025-39691"],"references":[{"type":"WEB","url":"https://nvd.nist.gov/vuln/detail/CVE-2025-39691"}],"affected":[{"package":{"name":"kernel","ecosystem":"Azure Linux:2","purl":"pkg:rpm/azure-linux/kernel"},"ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0"},{"fixed":"5.15.200.1-1"}]}],"database_specific":{"source":"https://github.com/microsoft/AzureLinuxVulnerabilityData/blob/main/osv/AZL-73710.json"}}],"schema_version":"1.7.5"}