Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
IronPort-SDR: gbP/ext6zml+5mzPNIFnUxdm9RhjXGaN5d8MmzLT42hObogu7SfOhHMIbuB6rD8qbLVInZDSN/
 V+lPuyFNt6UosMQ7ZUblqU/cC1+MLwYiIcpIDJ84nYDm/scPZcVPfCr8v2MvwFx9MGoobRO9So
 Gef/MJmAPobhaAI1u6SP4ZVfG7TePq5hq3qoIiUWNvvgeT/lpEhaFeDLSEKTpQOUa92m6Hzimq
 sShMTelcovIFO9FZzrGA9IhBId7/lhzzj3S3U9FKksuWXddOBRTfoEFz8CFv6AL2pzlf2DhkEO
 USk=
IronPort-SDR: RlkxMf15PfUxIYrYgy1TGCYPJhh3iZuC2s71tqg8eemlbHwjXIz11HK94X00fU9g7pXU7CNrjo
 iSiTol7TeXYSJMG720dJWT+YsnfSPXspdgIJZ+y672JeCoVQ2ythf8En7dFyoHHcKFNeygQzhn
 Z18FqBGJT2qlZHSBkYkoWcDdGc/AyraCB4ZMsSIqvC253gQUCHl78urBLAJKpIlcn/fi8P7ytw
 5joR++Gj0F8PkbHseXqruN0tuLx+mY4SoQWg6vsSV93bUEwSmoTsODibrbJQizlAy9kahnvJkq
 7oI=
WDCIronportException: Internal
Message-ID: <dc4e757a-737d-0bfa-c85d-9521feaa8d5f@opensource.wdc.com>
Date:   Tue, 15 Nov 2022 12:06:06 +0900
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.4.1
Subject: Re: [PATCH] pata_parport: add driver (PARIDE replacement)
Content-Language: en-US
To:     Ondrej Zary <linux@zary.sk>
Cc:     Christoph Hellwig <hch@lst.de>,
        Sergey Shtylyov <s.shtylyov@omp.ru>,
        Jens Axboe <axboe@kernel.dk>, Tim Waugh <tim@cyberelk.net>,
        linux-block@vger.kernel.org, linux-parport@lists.infradead.org,
        linux-ide@vger.kernel.org, linux-kernel@vger.kernel.org
References: <20220312144415.20010-1-linux@zary.sk>
 <202211140853.11115.linux@zary.sk>
 <f8ce8ecd-cadd-d9ca-d2fa-1251804344f0@opensource.wdc.com>
 <202211142025.46723.linux@zary.sk>
From:   Damien Le Moal <damien.lemoal@opensource.wdc.com>
Organization: Western Digital Research
In-Reply-To: <202211142025.46723.linux@zary.sk>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 11/15/22 04:25, Ondrej Zary wrote:
> On Monday 14 November 2022 09:03:28 Damien Le Moal wrote:
>> On 11/14/22 16:53, Ondrej Zary wrote:
>>> On Monday 14 November 2022, Damien Le Moal wrote:
>>>> On 11/12/22 20:17, Ondrej Zary wrote:
>>>>> On Wednesday 19 October 2022 09:34:31 Christoph Hellwig wrote:
>>>>>> It's been a while - did you get a chance to make some progress on
>>>>>> this?  Do you need any help to unblock you?
>>>>>>
>>>>>
>>>>> Sorry again, I'm back now. Trying to fix locking problems.
>>>>> Added this to each function for analysis how the functions are called wrt.
>>>>> locking:
>>>>>
>>>>> 	printk("%s, locked=%d\n", __FUNCTION__, spin_is_locked(ap->lock));
>>>>
>>>> Do you have your code somewhere that we can look at ?
>>>
>>> This is the current version with debug printks. I've also added dump_stack()
>>> to find out the code path but haven't analyzed the output yet.
>>
>> Can you send a proper patch ? Or a link to a git tree ? That is easier to
>> handle than pasted code in an email...
> 
> Patch against what? I don't have a git server.

patch against current 6.1-rc, or against an older kernel should be OK too.
But please "git send-email" a patch, or push your dev tree to github ?

> I've done some call trace analysis. These code paths are calling
> pata_parport functions with ap->lock locked during init.
> 
> Comm: kworker, Workqueue: ata_sff ata_sff_pio_task
> ata_sff_hsm_move -> ata_pio_sectors-> ata_sff_altstatus -> pata_parport_tf_read -> pata_parport_check_altstatus
> ata_sff_hsm_move -> ata_sff_altstatus -> pata_parport_tf_read -> pata_parport_check_altstatus
> ata_sff_pio_task -> ata_sff_busy_wait -> pata_parport_check_status
> ata_sff_hsm_move -> ata_wait_idle -> ata_sff_busy_wait -> pata_parport_check_status
> ata_sff_hsm_move -> ata_hsm_qc_complete -> ata_sff_irq_on -> ata_wait_idle -> ata_sff_busy_wait -> pata_parport_check_status
> ata_sff_pio_task -> ata_sff_hsm_move -> ata_pio_sectors -> ata_pio_sector -> ata_pio_xfer -> pata_parport_data_xfer
> ata_sff_pio_task -> ata_sff_hsm_move -> pata_parport_data_xfer
> ata_sff_pio_task -> ata_sff_hsm_move -> pata_parport_tf_read
> ata_sff_hsm_move -> ata_hsm_qc_complete -> ata_qc_complete -> fill_result_tf -> ata_sff_qc_fill_rtf -> pata_parport_tf_read
> ata_sff_hsm_move -> ata_pio_sectors -> ata_sff_altstatus -> pata_parport_check_altstatus
> ata_sff_hsm_move -> ata_sff_altstatus -> pata_parport_check_altstatus
> 
> Comm: modprobe
> ata_host_start -> ata_eh_freeze_port -> ata_sff_freeze -> pata_parport_check_status
> 
> Comm: scsi_eh_4
> ata_eh_recover -> ata_eh_reset -> ata_eh_thaw_port -> ata_sff_thaw -> ata_sff_irq_on -> ata_wait_idle -> ata_sff_busy_wait -> pata_parport_check_status
> ata_eh_reset -> ata_eh_freeze_port -> ata_sff_freeze -> pata_parport_check_status
> ata_scsi_error -> ata_scsi_port_error_handler -> ata_port_freeze -> ata_sff_freeze -> pata_parport_check_status
> ata_sff_error_handler -> pata_parport_drain_fifo -> pata_parport_check_status

What exactly are the issues you are having with ap->lock ? It looks like
you have done a lot of analysis of the code, but without any context about
the problem, I do not understand what I am looking at.

-- 
Damien Le Moal
Western Digital Research