git bisect replay produces wrong result

Neil Brown· Aug 30, 2015, 04:38 UTC · lore

Hi,
 the following git-bisect log - applied to a recent linux-kernel tree
 produced different end results when I use "git bisect replay"
 and when I just run it as a shell script.

$ git bisect replay /tmp/log 
We are not bisecting.
Bisecting: a merge base must be tested
[2decb2682f80759f631c8332f9a2a34a02150a03] Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net

$ bash /tmp/log
....
Bisecting: 2 revisions left to test after this (roughly 1 step)
[57127645d79d2e83e801f141f7d03f64accf28aa] s390/zcrypt: Introduce new SHA-512 based Pseudo Random Generator.

Is "git bisect replay" doing the wrong thing, or am I confused?

I tested on 2.5.0, and the current HEAD.

Thanks, NeilBrown

git bisect start # bad: [5ebe6afaf0057ac3eaeb98defd5456894b446d22] Linux 4.1-rc2 git bisect bad 5ebe6afaf0057ac3eaeb98defd5456894b446d22 # good: [39a8804455fb23f09157341d3ba7db6d7ae6ee76] Linux 4.0 git bisect good 39a8804455fb23f09157341d3ba7db6d7ae6ee76 # good: [6c373ca89399c5a3f7ef210ad8f63dc3437da345] Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next git bisect good 6c373ca89399c5a3f7ef210ad8f63dc3437da345 # good: [2c33ce009ca2389dbf0535d0672214d09738e35e] Merge Linus master into drm-next git bisect good 2c33ce009ca2389dbf0535d0672214d09738e35e # good: [7d2b6ef19cf0f98cef17aa5185de3631a618710a] Merge tag 'armsoc-drivers' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc git bisect good 7d2b6ef19cf0f98cef17aa5185de3631a618710a # good: [836ee4874e201a5907f9658fb2bf3527dd952d30] Merge tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux git bisect good 836ee4874e201a5907f9658fb2bf3527dd952d30 # good: [78d425677217b655ed36c492a070b5002832fc73] Merge tag 'platform-drivers-x86-v4.1-1' of git://git.infradead.org/users/dvhart/linux-platform-drivers-x86 git bisect good 78d425677217b655ed36c492a070b5002832fc73 # good: [39376ccb1968ba9f83e2a880a8bf02ad5dea44e1] Merge git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf git bisect good 39376ccb1968ba9f83e2a880a8bf02ad5dea44e1 # bad: [64887b6882de36069c18ef2d9623484d6db7cd3a] Merge branch 'for-linus-4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs git bisect bad 64887b6882de36069c18ef2d9623484d6db7cd3a # bad: [9dbbe3cfc3c208643cf0e81c8f660f43e1b4b2e8] Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm git bisect bad 9dbbe3cfc3c208643cf0e81c8f660f43e1b4b2e8 # bad: [dcca8de0aa597f14e31a1b38690626c9f6745fd5] Merge tag 'usb-4.1-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb git bisect bad dcca8de0aa597f14e31a1b38690626c9f6745fd5 # bad: [3d99e3fe13d473ac4578c37f477a59b829530764] Merge branch 'stable' of git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile git bisect bad 3d99e3fe13d473ac4578c37f477a59b829530764 # good: [b7d14f3a92223c3f5e52e9f20c74cb96dc130e87] s390/mm: correct transfer of dirty & young bits in __pmd_to_pte git bisect good b7d14f3a92223c3f5e52e9f20c74cb96dc130e87

Christian Couder· Aug 31, 2015, 07:54 UTC · re: Neil Brown · lore

Re: git bisect replay produces wrong result

Hi,

On Sun, Aug 30, 2015 at 6:38 AM, Neil Brown <neil@brown.name> wrote:

Show 18 quoted lines

>
> Hi,
>  the following git-bisect log - applied to a recent linux-kernel tree
>  produced different end results when I use "git bisect replay"
>  and when I just run it as a shell script.
>
> $ git bisect replay /tmp/log
> We are not bisecting.
> Bisecting: a merge base must be tested
> [2decb2682f80759f631c8332f9a2a34a02150a03] Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
>
>
> $ bash /tmp/log
> ....
> Bisecting: 2 revisions left to test after this (roughly 1 step)
> [57127645d79d2e83e801f141f7d03f64accf28aa] s390/zcrypt: Introduce new SHA-512 based Pseudo Random Generator.
>
> Is "git bisect replay" doing the wrong thing, or am I confused?

I think you are right and "git bisect replay" is buggy. "git bisect replay" should display the same thing. I will take a look at fixing this.

Thanks for the report, Christian.