On Tue, Jan 27, 2015 at 10:44 PM, Yinghai Lu <yinghai@xxxxxxxxxx> wrote: > On Tue, Jan 27, 2015 at 8:02 PM, Bjorn Helgaas <bhelgaas@xxxxxxxxxx> wrote: >> Would you mind opening a report at http://bugzilla.kernel.org and >> attaching the complete dmesg logs from both kernels to it? If there >> was no working kernel, collect a log from v3.19-rc6 with >> "pci=realloc=off" instead. > > would be better if with "debug ignore_loglevel pci=realloc=off". This is unrelated to the bug, and the parameters Yinghai suggested are fine, but just out of curiosity , I thought "debug" and "ignore_loglevel" only affected the output that goes to the console. "ignore_loglevel" is handled by ignore_loglevel_setup() in kernel/printk.c; it sets "ignore_loglevel = true". I think "debug" is handled by debug_kernel() in init/main.c; it sets "console_loglevel = CONSOLE_LOGLEVEL_DEBUG". But there are about a million "debug" occurrences in the kernel, so maybe it does more than just set console_loglevel. The net effect is that I thought: - The dmesg log always contains everything, regardless of the console_loglevel. - Using "ignore_loglevel" means everything goes to the console, in addition to being in dmesg, regardless of what console_loglevel is. So isn't it enough to just collect the dmesg log, without using "debug" or "ignore_loglevel"? Obviously, if the system doesn't boot, you need to collect a console log, and then you would need "ignore_loglevel". And if you use "ignore_loglevel", isn't that enough by itself? I wouldn't think using "debug" in addition would make any difference. Or maybe "debug" also turns on some dynamic debugging magic or something? If it does, that might be worth mentioning in kernel-parameters.txt. Bjorn -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
