On Tue, 9 Feb 2016, Yaroslav Molochko wrote: > After upgrade to 4.5.0-rc3 I've got following stack trace, never seen > that before: > ================================================================================ > Feb 08 22:58:56 x kernel: UBSAN: Undefined behaviour in > drivers/usb/host/ehci-hub.c:873:47 > Feb 08 22:58:56 x kernel: index -1 is out of range for type 'u32 [1]' > Feb 08 22:58:56 x kernel: CPU: 0 PID: 1 Comm: swapper/0 Not tainted > 4.5.0-rc3 #98 > Feb 08 22:58:56 x kernel: Hardware name: LENOVO 20BSCTO1WW/20BSCTO1WW, > BIOS N14ET32W (1.10 ) 08/13/2015 > Feb 08 22:58:56 x kernel: 0000000000000000 0000000084b750fa > ffff880224deb358 ffffffff81832ca4 > Feb 08 22:58:56 x kernel: ffffffffffffffff 0000000084b750fa > ffff880224deb380 ffffffffffffffff > Feb 08 22:58:56 x kernel: ffff880224deb370 ffffffff818a6360 > ffffffff833ea920 ffff880224deb3c8 > Feb 08 22:58:56 x kernel: Call Trace: > Feb 08 22:58:56 x kernel: [<ffffffff81832ca4>] dump_stack+0xaf/0x10c > Feb 08 22:58:56 x kernel: [<ffffffff818a6360>] ubsan_epilogue+0x14/0x56 > Feb 08 22:58:56 x kernel: [<ffffffff818a6c47>] > __ubsan_handle_out_of_bounds+0x86/0xb3 > Feb 08 22:58:56 x kernel: [<ffffffff81e65a88>] ehci_hub_control+0xcf/0x141e I have no idea why this didn't show up earlier. In any case it doesn't matter; the -1 index is part of an address calculation for a pointer that doesn't get used in this pathway. This seems to be a common problem with UBSAN. It spots invalid or questionable calculations in situations where they make no difference because the results don't get used. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
