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 for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <5004AD3A.7060701@gmail.com>
Date:	Tue, 17 Jul 2012 08:09:30 +0800
From:	Jiang Liu <liuj97@...il.com>
To:	Bjorn Helgaas <bhelgaas@...gle.com>
CC:	Jiang Liu <jiang.liu@...wei.com>, Don Dutile <ddutile@...hat.com>,
	Yinghai Lu <yinghai@...nel.org>,
	Taku Izumi <izumi.taku@...fujitsu.com>,
	"Rafael J . Wysocki" <rjw@...k.pl>,
	Kenji Kaneshige <kaneshige.kenji@...fujitsu.com>,
	Yijing Wang <wangyijing@...wei.com>,
	Keping Chen <chenkeping@...wei.com>,
	linux-kernel@...r.kernel.org, linux-pci@...r.kernel.org
Subject: Re: [RFC PATCH 05/14] PCI: add access functions for PCIe capabilities
 to hide PCIe spec differences

On 07/17/2012 01:29 AM, Bjorn Helgaas wrote:
> On Sun, Jul 15, 2012 at 10:47 AM, Jiang Liu <liuj97@...il.com> wrote:
>> On 07/13/2012 04:49 AM, Bjorn Helgaas wrote:
>>>> Hi Bjorn,
>>>>         It's a little risk to change these PCIe capabilities access
>>>> functions as void. On some platform with hardware error detecting/correcting
>>>> capabilities, such as EEH on Power, it would be better to return
>>>> error code if hardware error happens during accessing configuration registers.
>>>>         As I know, coming Intel Xeon processor may provide PCIe hardware
>>>> error detecting capability similar to EEH on power.
>>>
>>> I guess I'm playing devil's advocate here.  As a general rule, people
>>> don't check the return value of pci_read_config_*() or
>>> pci_write_config_*().  Unless you change them all, most callers of
>>> pci_pcie_capability_read_*() and _write_*() won't check the returns
>>> either.  So I'm not sure return values are an effective way to detect
>>> those hardware errors.
>>>
>>> How do these EEH errors get detected or reported today?  Do the
>>> drivers check every config access for success?  Adding those checks
>>> and figuring out how to handle errors at every possible point doesn't
>>> seem like a recipe for success.
>>
>> Hi Bjorn,
>>         Sorry for later reply, on travel these days.
>>         Yeah, it's true that most driver doesn't check return values of configuration
>> access functions, but there are still some drivers which do check return value of
>> pci_read_config_xxx(). For example, pciehp driver checks return value of CFG access
>> functions.
>>
>>         It's not realistic to enhance all drivers, but we may focus on a small set of
>> drivers for hardwares on specific high-end servers. For RAS features, we can never provide
>> perfect solutions, so we prefer some improvements. After all a small improvement is still
>> an improvement:)
>>
>>         I'm only familiar with PCI on IA64 and x86. For PowerPC, I just know that the OS
>> may query firmware whether there's some hardware faults if pci_cfg_read_xxx() returns
>> all 1s. For PCI on IA64, SAL may handle PCI hardware errors and return error code to
>> pci_cfg_read_xxx(). For x86, I think it will have some mechanisms to report hardware faults
>> like SAL on IA64.
>>
>>         So how about keeping consistence with pci_cfg_read_xxx() and pci_user_cfg_read_xxx()?
> 
> My goal is "the caller should never have to know whether this is a v1
> or v2 capability."  Returning any error other than one passed along
> from pci_read/write_config_xxx() means we miss that goal.  Perhaps the
> goal is unattainable, but I haven't been convinced yet.
> 
> I think hardware error detection is irrelevant to this discussion.
> After reading Documentation/PCI/pci-error-recovery.txt, I'm even less
> convinced that checking return values from pci_read/write_config_xxx()
> or pci_pcie_capability_read/write_xxx() is a useful way to detect
> hardware errors.
> 
> Having drivers detect hardware failures by checking for config access
> errors is neither necessary nor sufficient.  It's not necessary
> because a platform can implement a config accessor that checks *every*
> access and reports failures to the driver via the pci_error_handler
> framework.  It's not sufficient because config accesses are rare
> (usually only at init-time), and hardware failures may happen at
> arbitrary other times.
> 
> In my opinion, the only relevant question is whether a caller of
> pci_pcie_capability_read/write_xxx() needs to know whether a register
> is implemented (i.e., we have a v2 capability) or not.  For reads, I
> don't think there's a case where fabricating a value of zero when
> reading an unimplemented register is a problem.
> 
> Writes are obviously more interesting, but I'm still not sure there's
> a case where silently dropping a write to an unimplemented register is
> a problem.  The "capability" registers are read-only, so there's no
> problem if we drop writes to them.  The "status" registers are
> generally RO or RW1C, where it's only meaningful to write a non-zero
> value if you're previously *read* a non-zero value.  The "control"
> registers are often RW, of course, but generally it's only meaningful
> to write a non-zero value when a non-zero bit in the "capability"
> register has previously told you that something is supported.
Hi Bjorn,
	I'm convinced by you that we shouldn't return error when accessing
an unimplemented PCIe capabilities register and just hide the differences 
among V1/V2 specification. Then how about returning error from
"pci_read/write_config_xxx()" to callers of pci_pcie_capabilitiy_read/write_xxx()?
I still prefer to return error code to keep consistence with other configuration
space access interfaces:)
	Thanks!
	Gerry

> 
> Bjorn
> 


--
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/

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ