(arrow-dotnet) branch main updated: Use unsigned long integers for computing bit-length of fixed-width types via the CData ABI (#251)

[curth]

This is an automated email from the ASF dual-hosted git repository.

curth pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/arrow-dotnet.git


The following commit(s) were added to refs/heads/main by this push:
     new 56a9a5e  Use unsigned long integers for computing bit-length of 
fixed-width types via the CData ABI (#251)
56a9a5e is described below

commit 56a9a5e53e9fa52224a678e7f2e504bb4399de58
Author: Joshua Klein <mobiuskl...@gmail.com>
AuthorDate: Mon Feb 2 10:10:53 2026 -0500

    Use unsigned long integers for computing bit-length of fixed-width types 
via the CData ABI (#251)
    
    ## What's Changed
    
    Uses unsigned 64-bit integers to compute the intermediate bit length of
    an array to avoid arithmetic overflow.
    
    I could factorize this expression differently to avoid the casts, but I
    didn't think this was a sufficiently heavy code path that it'd need more
    attention.
    
    Closes #250 .
    
    ---------
    
    Co-authored-by: Adam Reeve <adre...@gmail.com>
---
 src/Apache.Arrow/C/CArrowArrayImporter.cs | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/src/Apache.Arrow/C/CArrowArrayImporter.cs 
b/src/Apache.Arrow/C/CArrowArrayImporter.cs
index c454380..2dc4adb 100644
--- a/src/Apache.Arrow/C/CArrowArrayImporter.cs
+++ b/src/Apache.Arrow/C/CArrowArrayImporter.cs
@@ -470,7 +470,10 @@ namespace Apache.Arrow.C
                 int length = checked((int)cArray->offset + 
(int)cArray->length);
                 int valuesLength;
                 if (bitWidth >= 8)
-                    valuesLength = checked(length * bitWidth / 8);
+                {
+                    Debug.Assert(bitWidth % 8 == 0);
+                    valuesLength = checked((int)((ulong)length * 
(ulong)bitWidth / 8ul));
+                }
                 else
                     valuesLength = 
checked((int)BitUtility.RoundUpToMultipleOf8(length) / 8);