linux-kernel - Re: [PATCH v4 00/15] PCI: let the core manage slot names

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] [day] [month] [year] [list]

Date:	Thu, 09 Oct 2008 14:01:53 +0900
From:	Kenji Kaneshige <kaneshige.kenji@...fujitsu.com>
To:	Alex Chiang <achiang@...com>,
	Kenji Kaneshige <kaneshige.kenji@...fujitsu.com>,
	jbarnes@...tuousgeek.org, kristen.c.accardi@...el.com,
	matthew@....cx, linux-pci@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH v4 00/15] PCI: let the core manage slot names

Alex Chiang wrote:
> * Kenji Kaneshige <kaneshige.kenji@...fujitsu.com>:
>> Hi Alex-san,
>>
>> I have some ideas and made some patches for comments I sent about
>> your [PATCH v4 02/15] and [PATCH v4 03/15]. Please take a look.
>> There are three patches.
> 
> Hi Kenji-san,
> 
> Thanks for doing this work.
> 
> I tested your patches, but found a problem with refcounting in
> your 02/03, and the slot directories in sysfs remained, even
> after rmmod of all drivers.
> 
> I decided that things are getting too complicated with all these
> new interfaces, so I got rid of them and updated the
> pci_create_slot API instead, to take a 'rename' parameter. That
> way, creating a slot and overriding its name can become an atomic
> operation.
> 
> That should make the race conditions go away, and the code is
> much easier to understand as well.
> 

OK.
I guess one of the thing that will bother us is how to handle
the case pci_slot created by pci_create_slot() already has its
hotplug callbacks in pci_hp_register(). Current code calls
pci_destroy_slot() and return -EBUSY in this case. With the
new API which takes 'rename' parameterIn addtion to that, we
need to rename the slot as it used to be, in addition.


>>  - [01/03] Sample patch for [PATCH v4 02/15]
> 
> This is a good patch by itself. I think you should submit it to
> Jesse. I did not need it after reworking to my new design.
> 
>>  - [02/03] Sample patch for [PATCH v4 03/15]
>>    NOTE:This doesn't target the comment about changing exported
>>         symbol name.
> 
> I had to stare at this patch for a long time to understand it,
> and finally saw that you were changing the rename logic to detect
> if we were trying to rename an existing slot.
> 
> Unfortunately, it had some problem with the refcounting, and in
> this scenario:
> 
> 	- pci_slot loaded
> 	- fakephp dup_slots=1 loaded
> 	- pci_slot unloaded
> 
> The slots claimed by pci_slot (but _not_ by fakephp) were never
> released.
> 
> By the time I got this far, I was already thinking about a
> redesign, so I did not try and debug further...
> 
>>  - [03/03] Sample patch for [PATCH v4 14/15]
>>    This is needed because above two patches make your [PATCH v4
>>    14/15] can not be applied.
> 
> Not needed, since I re-designed the approach.
> 
>> Note: I made those patches as replacement of your corresponding
>> ones. So those patches are NOT for applying on top your original
>> patches.
> 
> Again, thank you very much for all the review and hard work, and
> sorry for causing so much churn. :-/
> 
> I'll be sending out the new patch series shortly.
> 

No problem.
I'm looking forward to looking at new patches.

Thanks,
Kenji Kaneshige



--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/