CVE-2024-26798

Severity CVSS v4.0:
Pending analysis
Type:
Unavailable / Other
Publication date:
04/04/2024
Last modified:
06/02/2026

Description

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> fbcon: always restore the old font data in fbcon_do_set_font()<br /> <br /> Commit a5a923038d70 (fbdev: fbcon: Properly revert changes when<br /> vc_resize() failed) started restoring old font data upon failure (of<br /> vc_resize()). But it performs so only for user fonts. It means that the<br /> "system"/internal fonts are not restored at all. So in result, the very<br /> first call to fbcon_do_set_font() performs no restore at all upon<br /> failing vc_resize().<br /> <br /> This can be reproduced by Syzkaller to crash the system on the next<br /> invocation of font_get(). It&amp;#39;s rather hard to hit the allocation failure<br /> in vc_resize() on the first font_set(), but not impossible. Esp. if<br /> fault injection is used to aid the execution/failure. It was<br /> demonstrated by Sirius:<br /> BUG: unable to handle page fault for address: fffffffffffffff8<br /> #PF: supervisor read access in kernel mode<br /> #PF: error_code(0x0000) - not-present page<br /> PGD cb7b067 P4D cb7b067 PUD cb7d067 PMD 0<br /> Oops: 0000 [#1] PREEMPT SMP KASAN<br /> CPU: 1 PID: 8007 Comm: poc Not tainted 6.7.0-g9d1694dc91ce #20<br /> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014<br /> RIP: 0010:fbcon_get_font+0x229/0x800 drivers/video/fbdev/core/fbcon.c:2286<br /> Call Trace:<br /> <br /> con_font_get drivers/tty/vt/vt.c:4558 [inline]<br /> con_font_op+0x1fc/0xf20 drivers/tty/vt/vt.c:4673<br /> vt_k_ioctl drivers/tty/vt/vt_ioctl.c:474 [inline]<br /> vt_ioctl+0x632/0x2ec0 drivers/tty/vt/vt_ioctl.c:752<br /> tty_ioctl+0x6f8/0x1570 drivers/tty/tty_io.c:2803<br /> vfs_ioctl fs/ioctl.c:51 [inline]<br /> ...<br /> <br /> So restore the font data in any case, not only for user fonts. Note the<br /> later &amp;#39;if&amp;#39; is now protected by &amp;#39;old_userfont&amp;#39; and not &amp;#39;old_data&amp;#39; as the<br /> latter is always set now. (And it is supposed to be non-NULL. Otherwise<br /> we would see the bug above again.)

Vulnerable products and versions

CPE From Up to
cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:* 5.15.64 (including) 5.15.151 (excluding)
cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:* 6.0 (including) 6.1.81 (excluding)
cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:* 6.2 (including) 6.6.21 (excluding)
cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:* 6.7 (including) 6.7.9 (excluding)
cpe:2.3:o:linux:linux_kernel:6.8:rc1:*:*:*:*:*:*
cpe:2.3:o:linux:linux_kernel:6.8:rc2:*:*:*:*:*:*
cpe:2.3:o:linux:linux_kernel:6.8:rc3:*:*:*:*:*:*
cpe:2.3:o:linux:linux_kernel:6.8:rc4:*:*:*:*:*:*
cpe:2.3:o:linux:linux_kernel:6.8:rc5:*:*:*:*:*:*
cpe:2.3:o:linux:linux_kernel:6.8:rc6:*:*:*:*:*:*