[PATCH 1 of 1] revlog: fix partial read error introduced in 57a41c08feab (issue1672)

[Matt Mackall]

On Sat, 2009-05-23 at 09:50 +0000, Henrik Stuart wrote:
> # HG changeset patch
> # User Henrik Stuart <hg at hstuart.dk>
> # Date 1243071825 -7200
> # Node ID 8ac77e18bdbe3c93f56ee5230bc99290a0995af3
> # Parent  9f85da26050895c67a162d0d2097255402618017
> revlog: fix partial read error introduced in 57a41c08feab (issue1672)
> 
> If an inline revlog is larger than _prereadsize, the remainder of the
> revlog is never read, or if util.openhardlinks() evaluates to False,
> e.g. on Windows.
> 
> diff -r 9f85da260508 -r 8ac77e18bdbe mercurial/revlog.py
> --- a/mercurial/revlog.py	Sat May 23 00:24:00 2009 +0200
> +++ b/mercurial/revlog.py	Sat May 23 11:43:45 2009 +0200
> @@ -380,6 +380,8 @@
>              index[0] = e
>              return index, nodemap, None
>  
> +        if size > _prereadsize:
> +            data += fp.read()
>          # call the C implementation to parse the index data
>          index, nodemap, cache = parsers.parse_index(data, inline)
>          return index, nodemap, cache

I came up with exactly this fix this morning. I've since changed it to:

diff -r 9f85da260508 mercurial/revlog.py
--- a/mercurial/revlog.py	Sat May 23 00:24:00 2009 +0200
+++ b/mercurial/revlog.py	Sat May 23 10:59:01 2009 -0500
@@ -322,7 +322,7 @@
         index = []
         nodemap =  {nullid: nullrev}
         n = off = 0
-        if len(data) < _prereadsize:
+        if len(data) == _prereadsize:
             data += fp.read() # read the rest
         l = len(data)
         while off + s <= l:
@@ -362,23 +362,19 @@
         self.size = struct.calcsize(indexformatng)
 
     def parseindex(self, fp, data, inline):
-        try:
-            size = len(data)
-            if size == _prereadsize:
-                size = util.fstat(fp).st_size
-        except AttributeError:
-            size = 0
-
-        if util.openhardlinks() and not inline and size > _prereadsize:
-            # big index, let's parse it on demand
-            parser = lazyparser(fp, size)
-            index = lazyindex(parser)
-            nodemap = lazymap(parser)
-            e = list(index[0])
-            type = gettype(e[0])
-            e[0] = offset_type(0, type)
-            index[0] = e
-            return index, nodemap, None
+        if len(data) == _prereadsize:
+            if util.openhardlinks() and not inline:
+                # big index, let's parse it on demand
+                parser = lazyparser(fp, size)
+                index = lazyindex(parser)
+                nodemap = lazymap(parser)
+                e = list(index[0])
+                type = gettype(e[0])
+                e[0] = offset_type(0, type)
+                index[0] = e
+                return index, nodemap, None
+            else:
+                data += fp.read()
 
         # call the C implementation to parse the index data
         index, nodemap, cache = parsers.parse_index(data, inline)

-- 
http://selenic.com : development and support for Mercurial and Linux