[gpshead]

After a failed del b[:n], bytearray_setslice_linear() moved ob_start back, but bytearray_resize_storage() had already realigned the data and reset ob_start, so the bytearray then started n bytes before its buffer. The front-delete path now moves ob_start and the size before the shrinking resize and no longer restores anything, so a failed shrink leaves the bytes removed and the larger block in place, raising MemoryError as the non-front path already does. The checks in bytearray_resize_lock_held() are split from a new bytearray_resize_unchecked() so the slice code can call it directly without changing other callers, and the non-front failure path now writes the trailing null byte it was missing.

Fixes #160303

ANT-2026-QRZ6IDTC

• Issue: bytearray reads and writes outside its buffer after del b[:n] fails with MemoryError #160303

[cmaloney]

@vstinner recently made it so that failed allocations keep the pre-existing bytes (#157242). That means we should always have a valid bytes. Backported to 3.15 (3.14 has a different internal bytearray structure).

bytearray implements a fast delete at front by having an ob_start member which can be incremented. Also fast delete at end: Leave ob_alloc as is and call Py_SET_SIZE to set the new size. The "effective" bytes being smaller than the allocation is okay.

The bug report sounds like we're getting that pointer manipulation off? I would strongly prefer not allocating a new bytes here with PyBytes_FromStringAndSize

[gpshead]

Reworked to keep ob_start and avoid the extra allocation: the delete now moves ob_start and the size before the resize. A failed shrink leaves the bytes removed and the larger block in place, raising MemoryError as the non-front path already does.

... we could have a failed shrink not raise at all, since the larger block is valid? I still like the error myself. del implies someone was hoping to release memory. Silently not doing so is presumably only going to happen in a high memory pressure situation? Being silent would then just mean that something else fails soon after.