Patchwork D8097: rust-nodemap: accounting for dead blocks

login
register
mail settings
Submitter phabricator
Date Feb. 8, 2020, 5:45 p.m.
Message ID <differential-rev-PHID-DREV-i4fkklbgyhx4xqwi4qo2-req@mercurial-scm.org>
Download mbox | patch
Permalink /patch/45050/
State New
Headers show

Comments

phabricator - Feb. 8, 2020, 5:45 p.m.
marmoute created this revision.
Herald added subscribers: mercurial-devel, kevincox.
Herald added a reviewer: hg-reviewers.

REVISION SUMMARY
  By the very append-only nature of the `NodeTree`, inserting
  new blocks has the effect of making some of the older ones
  useless as they become unreachable.
  
  Therefore some automatic housekeeping will need to be provided.
  This is standard procedure in the word of databases, under names
  such as "repack" or "vacuum".
  
  The new `masked_readonly_blocks()` will provide callers with
  useful information to decide if the nodetree is ripe for
  repacking, but all the `NodeTree` can provide is how many
  blocks have been masked in the currently mutable part. Analysing
  the readonly part would be way too long to do it for each
  transaction and defeat the whole purpose of nodemap persistence.
  Serializing callers (from the Python layer) will get this figure
  before each extraction and maintain an aggregate counter of
  unreachable blocks separately.
  
  Note: at this point, the most efficient repacking is just to restart
  afresh with a full rescan.

REPOSITORY
  rHG Mercurial

REVISION DETAIL
  https://phab.mercurial-scm.org/D8097

AFFECTED FILES
  rust/hg-core/src/revlog/nodemap.rs

CHANGE DETAILS




To: marmoute, #hg-reviewers
Cc: kevincox, mercurial-devel

Patch

diff --git a/rust/hg-core/src/revlog/nodemap.rs b/rust/hg-core/src/revlog/nodemap.rs
--- a/rust/hg-core/src/revlog/nodemap.rs
+++ b/rust/hg-core/src/revlog/nodemap.rs
@@ -254,6 +254,7 @@ 
     readonly: Box<dyn Deref<Target = [Block]> + Send>,
     growable: Vec<Block>,
     root: Block,
+    masked_inner_blocks: usize,
 }
 
 impl Index<usize> for NodeTree {
@@ -330,6 +331,7 @@ 
             readonly: readonly,
             growable: Vec::new(),
             root: root,
+            masked_inner_blocks: 0,
         }
     }
 
@@ -452,6 +454,7 @@ 
         let ro_len = ro_blocks.len();
         let glen = self.growable.len();
         if idx < ro_len {
+            self.masked_inner_blocks += 1;
             // TODO OPTIM I think this makes two copies
             self.growable.push(ro_blocks[idx].clone());
             (glen + ro_len, &mut self.growable[glen], glen + 1)
@@ -540,6 +543,22 @@ 
         }
         Ok(())
     }
+
+    /// Return the number of blocks in the readonly part that are currently
+    /// masked in the mutable part.
+    ///
+    /// The `NodeTree` structure has no efficient way to know how many blocks
+    /// are already unreachable in the readonly part.
+    pub fn masked_readonly_blocks(&self) -> usize {
+        if let Some(readonly_root) = self.readonly.last() {
+            if readonly_root == &self.root {
+                return 0;
+            }
+        } else {
+            return 0;
+        }
+        self.masked_inner_blocks + 1
+    }
 }
 
 pub struct NodeTreeBytes {
@@ -818,6 +837,7 @@ 
             readonly: sample_nodetree().readonly,
             growable: vec![block![0: Rev(1), 5: Rev(3)]],
             root: block![0: Block(1), 1:Block(3), 12: Rev(2)],
+            masked_inner_blocks: 1,
         };
         assert_eq!(nt.find_hex(&idx, "10")?, Some(1));
         assert_eq!(nt.find_hex(&idx, "c")?, Some(2));
@@ -826,6 +846,7 @@ 
         assert_eq!(nt.find_hex(&idx, "000")?, Some(NULL_REVISION));
         assert_eq!(nt.unique_prefix_len_hex(&idx, "000")?, Some(3));
         assert_eq!(nt.find_hex(&idx, "01")?, Some(9));
+        assert_eq!(nt.masked_readonly_blocks(), 2);
         Ok(())
     }
 
@@ -915,6 +936,8 @@ 
         assert_eq!(idx.find_hex("1a345")?, Some(3));
         assert_eq!(idx.find_hex("1a341")?, None);
 
+        // there's no readonly block to mask
+        assert_eq!(idx.nt.masked_readonly_blocks(), 0);
         Ok(())
     }
 
@@ -976,6 +999,8 @@ 
         assert_eq!(idx.find_hex("1235")?, Some(1));
         assert_eq!(idx.find_hex("131")?, Some(2));
         assert_eq!(idx.find_hex("cafe")?, Some(3));
+        // we did not add anything since init from readonly
+        assert_eq!(idx.nt.masked_readonly_blocks(), 0);
 
         idx.insert(4, "123A")?;
         assert_eq!(idx.find_hex("1234")?, Some(0));
@@ -983,12 +1008,18 @@ 
         assert_eq!(idx.find_hex("131")?, Some(2));
         assert_eq!(idx.find_hex("cafe")?, Some(3));
         assert_eq!(idx.find_hex("123A")?, Some(4));
+        // we masked blocks for all prefixes of "123", including the root
+        assert_eq!(idx.nt.masked_readonly_blocks(), 4);
 
+        eprintln!("{:?}", idx.nt);
         idx.insert(5, "c0")?;
         assert_eq!(idx.find_hex("cafe")?, Some(3));
         assert_eq!(idx.find_hex("c0")?, Some(5));
         assert_eq!(idx.find_hex("c1")?, None);
         assert_eq!(idx.find_hex("1234")?, Some(0));
+        // inserting "c0" is just splitting the 'c' slot of the mutable root,
+        // it doesn't mask anything
+        assert_eq!(idx.nt.masked_readonly_blocks(), 4);
 
         Ok(())
     }