Re: [PATCH v2] Running with `--enable-ubsan` leads to a qtest failure:

[Laurent Vivier <lvivier@redhat.com>]

CC Trivial

On 29/04/2025 17:56, Nabih Estefan wrote:
> v2: used ldl_le_p and lduw_l_p instead of memcpy as per upstream
> suggestion.
> 
> ```
> ../tests/qtest/libqos/igb.c:106:5: runtime error: load of misaligned address 0x562040be8e33 for type 'uint32_t', which requires 4 byte alignment
> ```
> Instead of straight casting the uint8_t array, we use memcpy to assure
> alignment is correct against uint32_t and uint16_t.
> 
> Signed-off-by: Nabih Estefan <nabihestefan@google.com>
> ---
>   tests/qtest/libqos/igb.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/tests/qtest/libqos/igb.c b/tests/qtest/libqos/igb.c
> index f40c4ec4cd..2e0bb58617 100644
> --- a/tests/qtest/libqos/igb.c
> +++ b/tests/qtest/libqos/igb.c
> @@ -104,10 +104,10 @@ static void igb_pci_start_hw(QOSGraphObject *obj)
>       e1000e_macreg_write(&d->e1000e, E1000_RDT(0), 0);
>       e1000e_macreg_write(&d->e1000e, E1000_RDH(0), 0);
>       e1000e_macreg_write(&d->e1000e, E1000_RA,
> -                        le32_to_cpu(*(uint32_t *)address));
> +                        ldl_le_p((uint32_t *)address));
>       e1000e_macreg_write(&d->e1000e, E1000_RA + 4,
>                           E1000_RAH_AV | E1000_RAH_POOL_1 |
> -                        le16_to_cpu(*(uint16_t *)(address + 4)));
> +                        lduw_le_p((uint16_t *)(address + 4)));
>   
>       /* Set supported receive descriptor mode */
>       e1000e_macreg_write(&d->e1000e,

Reviewed-by: Laurent Vivier <lvivier@redhat.com>
Tested-by: Laurent Vivier <lvivier@redhat.com>