lib/w64devkit
1
0
Fork 0
mirror of https://github.com/skeeto/w64devkit.git synced 2026-10-23 06:08:56 +03:00
Code Activity

ninja: Fix deps log recovery

Browse Source 
In DepsLog::Load(), the file offset used for truncation-based recovery
was advanced past each record immediately after reading its bytes from
disk, but before validating the record's semantic correctness (e.g. the
path checksum or node-id range checks). When a record was structurally
complete but semantically invalid, recovery truncated the file to an
offset that included the bad record, so it was never removed.

This created an infinite cycle: every build would load the deps log, hit
the corrupt record, "recover" by truncating everything after it (but not
the record itself), then append new dependency records that would be
destroyed by the next build's recovery. The result was spurious rebuilds
on every invocation.

Move the offset update to the end of the loop body, after all validation
has passed. Now when a record fails validation, the offset still points
before it, and truncation correctly removes it.

(This patch has been submitted upstream.)
This commit is contained in:
Christopher Wellons
2026-04-11 16:24:27 -04:00
parent dc1b730fc1
commit b010e46fba
2 changed files with 32 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
Dockerfile
View File

@@ -721,9 +721,11 @@ RUN $ARCH-gcc -DEXE=ccache.exe -DCMD=gcc \
FROM cross AS build-ninja
COPY --from=dl-ninja /dl/ /dl/
COPY src/ninja-*.patch $PREFIX/src/
WORKDIR /ninja
RUN cmake -DCMAKE_BUILD_TYPE=Release \
RUN cat $PREFIX/src/ninja-*.patch | patch -d/dl/ninja -p1 \
&& cmake -DCMAKE_BUILD_TYPE=Release \
-DCMAKE_SYSTEM_NAME=Windows \
-DCMAKE_CXX_COMPILER=$ARCH-g++ \
-DCMAKE_EXE_LINKER_FLAGS="-s" \

29
src/ninja-fix-recovery.patch Normal file
View File

@@ -0,0 +1,29 @@
Fix deps log recovery failing to remove corrupt records
In DepsLog::Load(), the file offset used for truncation-based recovery
was advanced past each record immediately after reading its bytes from
disk, but before validating the record's semantic correctness (e.g. the
path checksum or node-id range checks). When a record was structurally
complete but semantically invalid, recovery truncated the file to an
offset that included the bad record, so it was never removed.
This created an infinite cycle: every build would load the deps log, hit
the corrupt record, "recover" by truncating everything after it (but not
the record itself), then append new dependency records that would be
destroyed by the next build's recovery. The result was spurious rebuilds
on every invocation.
Move the offset update to the end of the loop body, after all validation
has passed. Now when a record fails validation, the offset still points
before it, and truncation correctly removes it.
--- a/src/deps_log.cc
+++ b/src/deps_log.cc
@@ -205,3 +205,2 @@ LoadStatus DepsLog::Load(const string& path, State* state, string* err) {
}
- offset += size + sizeof(size);
@@ -270,2 +269,3 @@ LoadStatus DepsLog::Load(const string& path, State* state, string* err) {
}
+ offset += size + sizeof(size);
}