Message ID | 20210511003845.2429846-1-swboyd@chromium.org (mailing list archive) |
---|---|
Headers | show |
Series | Add build ID to stacktraces | expand |
On Mon 2021-05-10 17:38:32, Stephen Boyd wrote: > This series adds the kernel's build ID[1] to the stacktrace header > printed in oops messages, warnings, etc. and the build ID for any module > that appears in the stacktrace after the module name. The goal is to > make the stacktrace more self-contained and descriptive by including the > relevant build IDs in the kernel logs when something goes wrong. This > can be used by post processing tools like script/decode_stacktrace.sh > and kernel developers to easily locate the debug info associated with a > kernel crash and line up what line and file things started falling apart > at. The entire series looks good to me. I reviewed carefully only the 5th patch touching printk/kallsyms/module code. I just scanned over the other patches touching kernel code and did not notice any obvious problem. I did not check the changes in decode_stacktrace.sh at all. I tried to get stacktraces on x86_64 and it worked as expected. Best Regards, Petr
From: Stephen Boyd > Sent: 11 May 2021 01:39 > > This series adds the kernel's build ID[1] to the stacktrace header > printed in oops messages, warnings, etc. and the build ID for any module > that appears in the stacktrace after the module name. The goal is to > make the stacktrace more self-contained and descriptive by including the > relevant build IDs in the kernel logs when something goes wrong. This > can be used by post processing tools like script/decode_stacktrace.sh > and kernel developers to easily locate the debug info associated with a > kernel crash and line up what line and file things started falling apart > at. > > To show how this can be used I've included a patch to > decode_stacktrace.sh that downloads the debuginfo from a debuginfod > server. ... > Here's an example lkdtm stacktrace on arm64. > > WARNING: CPU: 4 PID: 3255 at drivers/misc/lkdtm/bugs.c:83 lkdtm_WARNING+0x28/0x30 [lkdtm] > Modules linked in: lkdtm rfcomm algif_hash algif_skcipher af_alg xt_cgroup uinput xt_MASQUERADE > CPU: 4 PID: 3255 Comm: bash Not tainted 5.11 #3 aa23f7a1231c229de205662d5a9e0d4c580f19a1 > Hardware name: Google Lazor (rev3+) with KB Backlight (DT) ... > x1 : ffffff93fef15788 x0 : ffffffe3622352e0 > Call trace: > lkdtm_WARNING+0x28/0x30 [lkdtm ed5019fdf5e53be37cb1ba7899292d7e143b259e] > direct_entry+0x16c/0x1b4 [lkdtm ed5019fdf5e53be37cb1ba7899292d7e143b259e] > full_proxy_write+0x74/0xa4 Is there any way to get it to print each module ID only once? David - Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
On Tue, 11 May 2021 12:36:06 +0000 David Laight <David.Laight@ACULAB.COM> wrote: > > x1 : ffffff93fef15788 x0 : ffffffe3622352e0 > > Call trace: > > lkdtm_WARNING+0x28/0x30 [lkdtm ed5019fdf5e53be37cb1ba7899292d7e143b259e] > > direct_entry+0x16c/0x1b4 [lkdtm ed5019fdf5e53be37cb1ba7899292d7e143b259e] > > full_proxy_write+0x74/0xa4 > > Is there any way to get it to print each module ID only once? If there's a trivial way to do that, then perhaps it should be done, but for now, this patch series isn't as obnoxious as the previous versions. It only affects stack traces, and I'm fine with that. -- Steve
From: Steven Rostedt > Sent: 11 May 2021 13:53 > > On Tue, 11 May 2021 12:36:06 +0000 > David Laight <David.Laight@ACULAB.COM> wrote: > > > > x1 : ffffff93fef15788 x0 : ffffffe3622352e0 > > > Call trace: > > > lkdtm_WARNING+0x28/0x30 [lkdtm ed5019fdf5e53be37cb1ba7899292d7e143b259e] > > > direct_entry+0x16c/0x1b4 [lkdtm ed5019fdf5e53be37cb1ba7899292d7e143b259e] > > > full_proxy_write+0x74/0xa4 > > > > Is there any way to get it to print each module ID only once? > > If there's a trivial way to do that, then perhaps it should be done, but for > now, this patch series isn't as obnoxious as the previous versions. It only > affects stack traces, and I'm fine with that. True. Printing the id in the module list was horrid. The real downside is all the extra text that will overflow the in-kernel buffer. At least it shouldn't be extra lines causing screen wrap. Unless the variable names are long - hi rust :-) David - Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
On Tue 2021-05-11 12:58:47, David Laight wrote: > From: Steven Rostedt > > Sent: 11 May 2021 13:53 > > > > On Tue, 11 May 2021 12:36:06 +0000 > > David Laight <David.Laight@ACULAB.COM> wrote: > > > > > > x1 : ffffff93fef15788 x0 : ffffffe3622352e0 > > > > Call trace: > > > > lkdtm_WARNING+0x28/0x30 [lkdtm ed5019fdf5e53be37cb1ba7899292d7e143b259e] > > > > direct_entry+0x16c/0x1b4 [lkdtm ed5019fdf5e53be37cb1ba7899292d7e143b259e] > > > > full_proxy_write+0x74/0xa4 > > > > > > Is there any way to get it to print each module ID only once? > > > > If there's a trivial way to do that, then perhaps it should be done, but for > > now, this patch series isn't as obnoxious as the previous versions. It only > > affects stack traces, and I'm fine with that. > > True. Printing the id in the module list was horrid. > > The real downside is all the extra text that will overflow the > in-kernel buffer. > At least it shouldn't be extra lines causing screen wrap. > Unless the variable names are long - hi rust :-) Note that the ID is printed only when CONFIG_STACKTRACE_BUILD_ID is enabled. It will be used only by some distros/vendors that use it to download the debuginfo packages. Best Regards, Petr
From: Petr Mladek > Sent: 11 May 2021 15:22 > > On Tue 2021-05-11 12:58:47, David Laight wrote: > > From: Steven Rostedt > > > Sent: 11 May 2021 13:53 > > > > > > On Tue, 11 May 2021 12:36:06 +0000 > > > David Laight <David.Laight@ACULAB.COM> wrote: > > > > > > > > x1 : ffffff93fef15788 x0 : ffffffe3622352e0 > > > > > Call trace: > > > > > lkdtm_WARNING+0x28/0x30 [lkdtm ed5019fdf5e53be37cb1ba7899292d7e143b259e] > > > > > direct_entry+0x16c/0x1b4 [lkdtm ed5019fdf5e53be37cb1ba7899292d7e143b259e] > > > > > full_proxy_write+0x74/0xa4 > > > > > > > > Is there any way to get it to print each module ID only once? > > > > > > If there's a trivial way to do that, then perhaps it should be done, but for > > > now, this patch series isn't as obnoxious as the previous versions. It only > > > affects stack traces, and I'm fine with that. > > > > True. Printing the id in the module list was horrid. > > > > The real downside is all the extra text that will overflow the > > in-kernel buffer. > > At least it shouldn't be extra lines causing screen wrap. > > Unless the variable names are long - hi rust :-) > > Note that the ID is printed only when CONFIG_STACKTRACE_BUILD_ID > is enabled. It will be used only by some distros/vendors that > use it to download the debuginfo packages. Until Ubuntu decide to turn it on :-) Actually, for the use case, the id could be trimmed significantly. It is only trying to differentiate between builds of a specific module. So even 8 digits would be plenty. David - Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
On Tue, May 11, 2021 at 02:31:38PM +0000, David Laight wrote: > Actually, for the use case, the id could be trimmed significantly. > It is only trying to differentiate between builds of a specific module. > So even 8 digits would be plenty. asked and answered. please review the bidding.