Re: Regression with a0747a859ef6 ("bdi: add error handle for bdi_debug_register")

[Bruno Wolff III]

On Thu, Dec 21, 2017 at 22:01:33 +0800,
 weiping zhang <zwp10758@xxxxxxxxx> wrote:
> Hi,
> how do you do bisect ?build all kernel commit one by one ?
> as you did before:
> https://bugzilla.redhat.com/show_bug.cgi?id=1520982

I just did the one bisect using Linus' tree. After each build, I would do 
a test boot and see if the boot was normal or if I got errors and an 
eventual hang before boot.

Since then I have used git revert to revert just the problem commit from 
later kernels (such as v4.15-rc4) and when I do the system boots normally. 
And when I don't do the revert or just use stock Fedora kernels the problem 
occurs every time.

I also did a couple of tests with Josh Boyer's Fedora kernel tree that 
has Fedora patches on top of the development kernel.

> what kernel source code do you use that occur warning at device_add_disk?
> from fedora or any official release ? if so ,could you provide web link?

That was from an offical Fedora kernel. I believe I got it from the 
nodebug repo, but that kernel should be the same as the one that was 
normally used for rawhide. It is at 
https://koji.fedoraproject.org/koji/buildinfo?buildID=1007500 
but I don't know how much longer the binaries will stay available in koji. 

> if you use same kernel source code and same .config, why your own build
> Cann't trigger that warning ?

I don't know. The install script may build the initramfs differently. As 
far as I can tell, if the WARN_ON was triggered, I should have gotten 
output.