On Wed, Nov 8, 2017 at 9:12 AM, Fengguang Wu <fengguang...@intel.com> wrote:
>
> OK. Here is the original faddr2line output:
>
> $ ~/linux/scripts/faddr2line vmlinux vlan_device_event+0x7f5/0xa40
> vlan_device_event+0x7f5/0xa40:
> vlan_device_event at net/8021q/vlan.h:60
Hmm. Yes, that's not what I hoped for.
> I notice that this trace shows no additional inline files at all.
> Is it because I did some kconfig option wrong, so that inline info is
> lost? Eg.
>
> CONFIG_OPTIMIZE_INLINING=y (reading lib/Kconfig.debug, it looks better set
> to N)
> CONFIG_DEBUG_INFO_REDUCED=y
> CONFIG_DEBUG_INFO_SPLIT=y
Hmm. It might also be a compiler/linker issue. Debug info is sadly
often a second class citizen, because it's generally not something
that gets tested nearly as much as the actual _code_ that a compiler
generates.
So we've certainly had issues before with incomplete debug info.
That said, I'm not sure how well that CONFIG_DEBUG_INFO_SPLIT works,
I've only ever tested with
CONFIG_DEBUG_INFO=y
CONFIG_DEBUG_INFO_REDUCED=y
# CONFIG_DEBUG_INFO_SPLIT is not set
# CONFIG_DEBUG_INFO_DWARF4 is not set
which was the config that my scheduler example was generated from.
It might be an issue with the particular version of 'addr2line' too,
for all I know. The one I have is
GNU addr2line version 2.27-24.fc26
and who knows what else could influence it. I *thought* that as long
as you just had DEBUG_INFO enabled, it would automatically be ok, but
I was clearly wrong.
> [ 745.719623] BUG: unable to handle kernel paging request at 6b6b6f4f
> [ 745.732871] IP: vlan_device_event+0x7f5/0xa40:
> vlan_device_event at net/8021q/vlan.h:60
Ok, at least now there's no lost information, even if it doesn't have
that nice inlining information that I was hoping for.
Linus
