[Webkit-unassigned] [Bug 210345] New: [GTK][WPE] lowWatermarkPages() in MemoryPressureMonitor.cpp only searchs the "low" value inside the first "Node" section

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=210345

            Bug ID: 210345
           Summary: [GTK][WPE] lowWatermarkPages() in
                    MemoryPressureMonitor.cpp only searchs the "low" value
                    inside the first "Node" section
           Product: WebKit
           Version: WebKit Nightly Build
          Hardware: Unspecified
                OS: Linux
            Status: NEW
          Severity: Normal
          Priority: P2
         Component: WebKitGTK
          Assignee: webkit-unassigned at lists.webkit.org
          Reporter: psaavedra at igalia.com
                CC: aperez at igalia.com, bugs-noreply at webkitgtk.org
        Depends on: 209942

Related to https://bugs.webkit.org/show_bug.cgi?id=209942
See details in https://bugs.webkit.org/show_bug.cgi?id=209942#c18


The lowWatermarkPages() in Source/WebKit/UIProcess/linux/MemoryPressureMonitor.cpp is only searching for the “low” value inside the first “Node” section, without taking into account the kind of the memory zone.

Shouldn't we be searching for the “Normal” zone node, instead of the first one seen? In my laptop here the first zone is a DMA memory area:

  Node 0, zone      DMA
    per-node stats
        nr_inactive_anon 277959
        nr_active_anon 863174
        ...
    pages free      3951
          min       13 
          low       16
          high      19
          ...

… and picking “16” as the “low” value seems wrong. We will want to fix this to find the node which describes the “Normal” zone:

  Node 0, zone     Normal
    pages free     252746
          min      14907
          low      26572
          high     31069
          ...

Related question: What if there are more multiple nodes which describe nodes of “Normal” type? Should the “low” values be added together in that case?


Referenced Bugs:

https://bugs.webkit.org/show_bug.cgi?id=209942
[Bug 209942] [GTK][WPE] Replace fopen/fclose by fopen/fseek functions in MemoryPressureMonitor
-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20200410/50928134/attachment.htm>