sc/source/core/data/table1.cxx | 4 ++++
1 file changed, 4 insertions(+)
New commits:
commit dc73583573a31078c8408e820d749aaeab25bc32
Author: Noel Grandin <noel.gran...@collabora.co.uk>
AuthorDate: Fri May 31 15:39:17 2019 +0200
Commit: Noel Grandin <noel.gran...@collabora.co.uk>
CommitDate: Fri May 31 20:26:01 2019 +0200
crash-testing fdo86059-1.xlsx
blind fix since I can't reproduce. But I can see from the reported
threads that two different threads are entering SfxItemPool::Put and
consequently the debug runtime throws an iterator exception.
Change-Id: Idaa6a93b134edc49bda8cd7876bb216d5aec7037
Reviewed-on: https://gerrit.libreoffice.org/73272
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.gran...@collabora.co.uk>
diff --git a/sc/source/core/data/table1.cxx b/sc/source/core/data/table1.cxx
index be7d12b50150..b899bbb91cbf 100644
--- a/sc/source/core/data/table1.cxx
+++ b/sc/source/core/data/table1.cxx
@@ -2554,6 +2554,10 @@ ScColumnsRange ScTable::GetColumnsRange(SCCOL nColBegin,
SCCOL nColEnd) const
// out-of-line the cold part of the CreateColumnIfNotExists function
void ScTable::CreateColumnIfNotExistsImpl( const SCCOL nScCol ) const
{
+ // When doing multi-threaded load of, e.g. XLS files, we can hit this,
which calls
+ // into SfxItemPool::Put, in parallel with other code that calls into
SfxItemPool::Put,
+ // which is bad since that code is not thread-safe.
+ SolarMutexGuard aGuard;
const SCCOL aOldColSize = aCol.size();
aCol.resize( static_cast< size_t >( nScCol + 1 ) );
for (SCCOL i = aOldColSize; i <= nScCol; i++)
_______________________________________________
Libreoffice-commits mailing list
libreoffice-comm...@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/libreoffice-commits
