linux-kernel - Re: [PATCH v2] ata: libata-scsi: Fix get identity data failed

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <c4f4bc15-09b7-ffee-8e56-c857fc1431a1@huawei.com>
Date:   Sat, 6 May 2023 17:57:13 +0800
From:   yangxingui <yangxingui@...wei.com>
To:     John Garry <john.g.garry@...cle.com>,
        Damien Le Moal <dlemoal@...nel.org>, <jejb@...ux.ibm.com>,
        <martin.petersen@...cle.com>, <damien.lemoal@...nsource.wdc.com>
CC:     <linux-scsi@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
        <linuxarm@...wei.com>, <prime.zeng@...ilicon.com>,
        <kangfenglong@...wei.com>
Subject: Re: [PATCH v2] ata: libata-scsi: Fix get identity data failed



On 2023/5/5 17:51, John Garry wrote:
> On 05/05/2023 10:14, yangxingui wrote:
>>> hmmm... is it really acceptable that libata is referencing libsas? I 
>>> didn't think that it would be. libsas uses libata, not the other way 
>>> around.
>> Yeah, I didn't expect that either. Is there any other way? If so, is 
>> patch v1 OK?
> 
> I still think that we can do better than v1.
OK
> 
>>>
>>>>>   #include <linux/libata.h>
>>>>>   #include <linux/hdreg.h>
>>>>>   #include <linux/uaccess.h>
>>>>> @@ -2745,10 +2746,17 @@ static struct ata_device 
>>>>> *__ata_scsi_find_dev(struct ata_port *ap,
>>>>>    *    Associated ATA device, or %NULL if not found.
>>>>>    */
>>>>>   struct ata_device *
>>>>> -ata_scsi_find_dev(struct ata_port *ap, const struct scsi_device 
>>>>> *scsidev)
>>>> Why drop the const ?
>>>>
>>>>> +ata_scsi_find_dev(struct ata_port *ap, struct scsi_device *scsidev)
>>>>>   {
>>>>> -    struct ata_device *dev = __ata_scsi_find_dev(ap, scsidev);
>>>>> +    struct ata_device *dev;
>>>>> +
>>>>> +    if (ap->flags & ATA_FLAG_SAS_HOST) {
>>>
>>> And this is SAS host. Not necessarily libsas (even though with ipr 
>>> libata usage gone, it would be the only user).
>> Add a new flag only for libsas?
> 
> No, because of previous reason.
> 
> Please remind me - at what point do we error within ata_scsi_find_dev() 
> and return NULL for a libsas host?
The scsi_host_template filled by libsas and call ata_scsi_find_dev() has 
this problem. After sas_change_queue_depth() is fixed, only sas_ioctl() 
has this problem now.
> 
> Note: it would be good to include that commit message for future reference.
> 
> Maybe we could add a method to ata_port_operations to do this lookup. I 
> assume that is abusing ata_port_operations purpose, since it's mostly 
> for HW methods.
> 
> Or do we actually use sdev->hostdata for libata or libsas? If not, maybe 
> we could store the struct ata_device pointer there.
Well, it might be a way.

Thanks,
Xingui
.
> 
> I'm just thinking out loud now...
> 
> Thanks,
> John
> 
> 
> .