Re: Making type Datum be 8 bytes everywhere - Mailing list pgsql-hackers

Tomas Vondra <tomas@vondra.me> writes:
> While testing a different patch, I tried running with address sanitizer
> on rpi5, running the 32-bit OS (which AFAIK is 64-bit kernel and 32-bit
> user space). With that, stats_ext regression tests fail like this:

> extended_stats.c:1082:27: runtime error: store to misaligned address
> 0x036671dc for type 'Datum', which requires 8 byte alignment
> 0x036671dc: note: pointer points here
>   00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  00 00 00 00 00 00 7e
> 7f  08 00 00 00 7f 7f 7f 7f
>               ^

> This happens because build_sorted_items() does palloc(), and then
> accesses the pointer as array of structs, with a Datum field. And it
> apparently expects the pointer to be a multiple of 8 bytes. Isn't that a
> bit strange, with 32-bit user space? The pointer is indeed a multiple of
> 4B, so maybe the expected alignment is wrong?

I think build_sorted_items is plainly at fault here, where it does

    /* Compute the total amount of memory we need (both items and values). */
    len = data->numrows * sizeof(SortItem) + nvalues * (sizeof(Datum) + sizeof(bool));

    /* Allocate the memory and split it into the pieces. */
    ptr = palloc0(len);

    /* items to sort */
    items = (SortItem *) ptr;
    ptr += data->numrows * sizeof(SortItem);

    /* values and null flags */
    values = (Datum *) ptr;
    ptr += nvalues * sizeof(Datum);

This is silently assuming that sizeof(SortItem) is a multiple of
alignof(Datum), which on a 32-bit-pointer platform is not true
any longer.  We ought to MAXALIGN the two occurrences of
data->numrows * sizeof(SortItem).