"value":"In the Linux kernel, the following vulnerability has been resolved:\n\nUSB: core: Fix race by not overwriting udev->descriptor in hub_port_init()\n\nSyzbot reported an out-of-bounds read in sysfs.c:read_descriptors():\n\nBUG: KASAN: slab-out-of-bounds in read_descriptors+0x263/0x280 drivers/usb/core/sysfs.c:883\nRead of size 8 at addr ffff88801e78b8c8 by task udevd/5011\n\nCPU: 0 PID: 5011 Comm: udevd Not tainted 6.4.0-rc6-syzkaller-00195-g40f71e7cd3c6 #0\nHardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 05/27/2023\nCall Trace:\n <TASK>\n __dump_stack lib/dump_stack.c:88 [inline]\n dump_stack_lvl+0xd9/0x150 lib/dump_stack.c:106\n print_address_description.constprop.0+0x2c/0x3c0 mm/kasan/report.c:351\n print_report mm/kasan/report.c:462 [inline]\n kasan_report+0x11c/0x130 mm/kasan/report.c:572\n read_descriptors+0x263/0x280 drivers/usb/core/sysfs.c:883\n...\nAllocated by task 758:\n...\n __do_kmalloc_node mm/slab_common.c:966 [inline]\n __kmalloc+0x5e/0x190 mm/slab_common.c:979\n kmalloc include/linux/slab.h:563 [inline]\n kzalloc include/linux/slab.h:680 [inline]\n usb_get_configuration+0x1f7/0x5170 drivers/usb/core/config.c:887\n usb_enumerate_device drivers/usb/core/hub.c:2407 [inline]\n usb_new_device+0x12b0/0x19d0 drivers/usb/core/hub.c:2545\n\nAs analyzed by Khazhy Kumykov, the cause of this bug is a race between\nread_descriptors() and hub_port_init(): The first routine uses a field\nin udev->descriptor, not expecting it to change, while the second\noverwrites it.\n\nPrior to commit 45bf39f8df7f (\"USB: core: Don't hold device lock while\nreading the \"descriptors\" sysfs file\") this race couldn't occur,\nbecause the routines were mutually exclusive thanks to the device\nlocking. Removing that locking from read_descriptors() exposed it to\nthe race.\n\nThe best way to fix the bug is to keep hub_port_init() from changing\nudev->descriptor once udev has been initialized and registered.\nDrivers expect the descriptors stored in the kernel to be immutable;\nwe should not undermine this expectation. In fact, this change should\nhave been made long ago.\n\nSo now hub_port_init() will take an additional argument, specifying a\nbuffer in which to store the device descriptor it reads. (If udev has\nnot yet been initialized, the buffer pointer will be NULL and then\nhub_port_init() will store the device descriptor in udev as before.)\nThis eliminates the data race responsible for the out-of-bounds read.\n\nThe changes to hub_port_init() appear more extensive than they really\nare, because of indentation changes resulting from an attempt to avoid\nwriting to other parts of the usb_device structure after it has been\ninitialized. Similar changes should be made to the code that reads\nthe BOS descriptor, but that can be handled in a separate patch later\non. This patch is sufficient to fix the bug found by syzbot."
