Re: [PATCH] sata_fsl: Use struct_group() for memcpy() region

From: Damien Le Moal
Date: Thu Nov 18 2021 - 18:52:42 EST


On 2021/11/19 8:39, Kees Cook wrote:
> On Fri, Nov 19, 2021 at 08:17:14AM +0900, Damien Le Moal wrote:
>> On 2021/11/19 3:38, Kees Cook wrote:
>>> In preparation for FORTIFY_SOURCE performing compile-time and run-time
>>> field bounds checking for memcpy(), memmove(), and memset(), avoid
>>> intentionally writing across neighboring fields.
>>>
>>> Use struct_group() in struct command_desc around members acmd and fill,
>>> so they can be referenced together. This will allow memset(), memcpy(),
>>> and sizeof() to more easily reason about sizes, improve readability,
>>> and avoid future warnings about writing beyond the end of acmd:
>>>
>>> In function 'fortify_memset_chk',
>>> inlined from 'sata_fsl_qc_prep' at drivers/ata/sata_fsl.c:534:3:
>>> ./include/linux/fortify-string.h:199:4: warning: call to '__write_overflow_field' declared with attribute warning: detected write beyond size of field (1st parameter); maybe use struct_group()? [-Wattribute-warning]
>>> 199 | __write_overflow_field();
>>> | ^~~~~~~~~~~~~~~~~~~~~~~~
>>>
>>> Signed-off-by: Kees Cook <keescook@xxxxxxxxxxxx>
>>
>> This lacks some context with regard to FORTIFY_SOURCE and struct_group(). Is
>> that already in 5.16 ? It sounds like it is not. Do you want a ack ? Or do you
>> want me to queue this up for 5.17 ?
>
> Ah yes, some details are here in the earlier "big" series cover letter
> here:
> https://lore.kernel.org/linux-hardening/20210818060533.3569517-1-keescook@xxxxxxxxxxxx/
>
> One of the requests from earlier review was to split it up for separate
> trees for the maintainers that wanted to take stuff via their trees
> directly.
>
> The new helpers are landed as of v5.16-rc1, so it can go either way, but
> given that the merge window is closed, I would expect this to be for
> v5.17.
>
> I am happy to to carry it in my fortify topic branch that I'm expecting
> to send for 5.17, but totally up to you. Some folks like to take these
> changes via their trees, others would rather not be bothered with it. :)

OK. Since it looks like the compilation warning will trigger only when your big
series land in 5.17, I will queue this in for-5.17 (still need to create than
one). Is it ok with you ?

>
> Thanks!
>
> -Kees
>
>>
>> Cheers.
>>
>>> ---
>>> drivers/ata/sata_fsl.c | 10 ++++++----
>>> 1 file changed, 6 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/drivers/ata/sata_fsl.c b/drivers/ata/sata_fsl.c
>>> index e5838b23c9e0..fec3c9032606 100644
>>> --- a/drivers/ata/sata_fsl.c
>>> +++ b/drivers/ata/sata_fsl.c
>>> @@ -246,8 +246,10 @@ enum {
>>> struct command_desc {
>>> u8 cfis[8 * 4];
>>> u8 sfis[8 * 4];
>>> - u8 acmd[4 * 4];
>>> - u8 fill[4 * 4];
>>> + struct_group(cdb,
>>> + u8 acmd[4 * 4];
>>> + u8 fill[4 * 4];
>>> + );
>>> u32 prdt[SATA_FSL_MAX_PRD_DIRECT * 4];
>>> u32 prdt_indirect[(SATA_FSL_MAX_PRD - SATA_FSL_MAX_PRD_DIRECT) * 4];
>>> };
>>> @@ -531,8 +533,8 @@ static enum ata_completion_errors sata_fsl_qc_prep(struct ata_queued_cmd *qc)
>>> /* setup "ACMD - atapi command" in cmd. desc. if this is ATAPI cmd */
>>> if (ata_is_atapi(qc->tf.protocol)) {
>>> desc_info |= ATAPI_CMD;
>>> - memset((void *)&cd->acmd, 0, 32);
>>> - memcpy((void *)&cd->acmd, qc->cdb, qc->dev->cdb_len);
>>> + memset(&cd->cdb, 0, sizeof(cd->cdb));
>>> + memcpy(&cd->cdb, qc->cdb, qc->dev->cdb_len);
>>> }
>>>
>>> if (qc->flags & ATA_QCFLAG_DMAMAP)
>>>
>>
>>
>> --
>> Damien Le Moal
>> Western Digital Research
>


--
Damien Le Moal
Western Digital Research