"value":"In the Linux kernel, the following vulnerability has been resolved:\n\nata: pata_macio: Fix DMA table overflow\n\nKolbj\u00f8rn and Jon\u00e1\u0161 reported that their 32-bit PowerMacs were crashing\nin pata-macio since commit 09fe2bfa6b83 (\"ata: pata_macio: Fix\nmax_segment_size with PAGE_SIZE == 64K\").\n\nFor example:\n\n kernel BUG at drivers/ata/pata_macio.c:544!\n Oops: Exception in kernel mode, sig: 5 [#1]\n BE PAGE_SIZE=4K MMU=Hash SMP NR_CPUS=2 DEBUG_PAGEALLOC PowerMac\n ...\n NIP pata_macio_qc_prep+0xf4/0x190\n LR pata_macio_qc_prep+0xfc/0x190\n Call Trace:\n 0xc1421660 (unreliable)\n ata_qc_issue+0x14c/0x2d4\n __ata_scsi_queuecmd+0x200/0x53c\n ata_scsi_queuecmd+0x50/0xe0\n scsi_queue_rq+0x788/0xb1c\n __blk_mq_issue_directly+0x58/0xf4\n blk_mq_plug_issue_direct+0x8c/0x1b4\n blk_mq_flush_plug_list.part.0+0x584/0x5e0\n __blk_flush_plug+0xf8/0x194\n __submit_bio+0x1b8/0x2e0\n submit_bio_noacct_nocheck+0x230/0x304\n btrfs_work_helper+0x200/0x338\n process_one_work+0x1a8/0x338\n worker_thread+0x364/0x4c0\n kthread+0x100/0x104\n start_kernel_thread+0x10/0x14\n\nThat commit increased max_segment_size to 64KB, with the justification\nthat the SCSI core was already using that size when PAGE_SIZE == 64KB,\nand that there was existing logic to split over-sized requests.\n\nHowever with a sufficiently large request, the splitting logic causes\neach sg to be split into two commands in the DMA table, leading to\noverflow of the DMA table, triggering the BUG_ON().\n\nWith default settings the bug doesn't trigger, because the request size\nis limited by max_sectors_kb == 1280, however max_sectors_kb can be\nincreased, and apparently some distros do that by default using udev\nrules.\n\nFix the bug for 4KB kernels by reverting to the old max_segment_size.\n\nFor 64KB kernels the sg_tablesize needs to be halved, to allow for the\npossibility that each sg will be split into two."