764019 - Remove our prefetch when READOPS=0 on windows

Status: RESOLVED FIXED

(Core :: General, defect)

Description

[(dormant account)]

bug 757215 shows that my heuristic a slight loss. We should focus any profiling/optimization on the case where we know prefetch is nuked for us, ie bug 692255.

Comment 1

[Mike Hommey [:glandium]]

Attachment: Only rely on the maintenance service to enable preload on Windows

I think the experiment we were talking about yesterday would be a separate bug.

Comment 2

[Mike Hommey [:glandium]]

Attachment: Only rely on the maintenance service to enable preload on Windows

Need to leave ioCounters there because it's used for *GLUESTARTUP_READ_OPS and *GLUESTARTUP_READ_TRANSFER.

Comment 3

[Brian R. Bondy [:bbondy]]

Comment on attachment 632636 [details] [diff] [review]
Only rely on the maintenance service to enable preload on Windows

Review of attachment 632636 [details] [diff] [review]:
-----------------------------------------------------------------

::: browser/app/nsBrowserApp.cpp
@@ -234,5 @@
>    if (IsPrefetchDisabledViaService()) {
>      preloadReason = PRELOAD_SERVICE;
> -  } else if ((gotCounters && !ioCounters.ReadOperationCount)) {
> -    preloadReason = PRELOAD_IOCOUNT;
> -  }

You need to keep this closing brace otherwise it won't compile.

Comment 4

[Mike Hommey [:glandium]]

Attachment: Only rely on the maintenance service to enable preload on Windows

(In reply to Brian R. Bondy [:bbondy] from comment #3)
> @@ -234,5 @@
> >    if (IsPrefetchDisabledViaService()) {
> >      preloadReason = PRELOAD_SERVICE;
> > -  } else if ((gotCounters && !ioCounters.ReadOperationCount)) {
> > -    preloadReason = PRELOAD_IOCOUNT;
> > -  }
> 
> You need to keep this closing brace otherwise it won't compile.

Yeah, try just told me so :)

Comment 5

[Mike Hommey [:glandium]]

https://hg.mozilla.org/mozilla-central/rev/da76a072b2b0