New issue
Advanced search Search tips

Issue 827291 link

Starred by 1 user

Issue metadata

Status: Untriaged
Owner: ----
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 1
Type: Bug



Sign in to add a comment

Windows compile not writing out at least one object.

Project Member Reported by petermayo@chromium.org, Mar 29 2018

Issue description

https://logs.chromium.org/v/?s=chromium%2Fbb%2Fchromium.win%2FWin_Builder%2F52898%2F%2B%2Frecipes%2Fsteps%2Fcompile_confirm_no-op%2F0%2Fstdout


re-running ninja complains that:
ninja explain: output obj/content/browser/browser/render_widget_host_view_aura.obj older than most recent input ../../ui/aura/window.h (544013529 vs 544053179)
ninja explain: obj/content/browser/browser/render_widget_host_view_aura.obj is dirty
ninja explain: obj/content/browser/browser/render_widget_host_view_aura.obj is dirty

but the compile step output:
[3430/11262] CXX obj/content/browser/browser/render_widget_host_view_aura.obj


i.e. relatively early in the compile process, and ui/aura/window.h looks like a normal checkout out source file

it looks like something changed it, or the .obj failed to write.
 
Concurrently another windows builder failed like this

LINK : fatal error LNK1181: cannot open input file 'obj/content/browser/browser/render_widget_host_view_aura.obj'

Different slavename.  I wonder if something is causing/caused the current version to not get written out, but still let the compile step pass.
Cc: fdoray@chromium.org
Labels: -Pri-3 Pri-0
Summary: Windows compile not writing our at least one object. (was: recompile detected out of date object.)
this does not reproduce locally according to fdoray@

Is now repeatably blocking windows compiles on the tree.
Cc: osh...@chromium.org
Labels: -Pri-0 Pri-1
reverting a relatively innocuous CL has fixed this.
Summary: Windows compile not writing out at least one object. (was: Windows compile not writing our at least one object.)
Cc: petermayo@chromium.org
Re #3 is this fixed? Which CL was reverted?
Cc: -petermayo@chromium.org
It is not fixed in that there is no reason to suspect it will not happen again.
one instance was fixed by clearing the cache, and the other by reverting 

https://chromium-review.googlesource.com/c/chromium/src/+/983962

That CL does not seem to require extra dependencies, but caused failure.
the other was https://chromium-review.googlesource.com/c/chromium/src/+/984A831 


Again, the pri-0 -> pri-1 drop because this isn't actively affecting the tree status, but it has the potential to as soon as some otherwise innocuous CL lands after passing the CQ.  The root cause has not been found or fixed.
Components: -Infra Build
This doesn't sound like an infra issue, but rather a build configuration problem. Updating the component.
so is it ok to re-land crbug.com/984831 ?
sorry i meant the CL https://crrev.com/c/984831

Sign in to add a comment