linux-kernel - Re: [PATCH 2/3] GHES: Move memory initialisation to ghes

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]

Date:   Tue, 15 Aug 2017 11:10:05 +0100
From:   Punit Agrawal <punit.agrawal@....com>
To:     Borislav Petkov <bp@...e.de>
Cc:     linux-acpi@...r.kernel.org, lorenzo.pieralisi@....com,
        sudeep.holla@....com, linux-kernel@...r.kernel.org,
        "Rafael J . Wysocki" <rjw@...ysocki.net>,
        Punit Agrawal <punit.agrwal@....com>,
        James Morse <james.morse@....com>
Subject: Re: [PATCH 2/3] GHES: Move memory initialisation to ghes_probe()

Hi Boris,

Borislav Petkov <bp@...e.de> writes:

> On Tue, Aug 01, 2017 at 02:36:07PM +0100, Punit Agrawal wrote:
>> During the driver init, the ghes driver initialises some data structures
>
> Let's stick to "GHES" in capital letters everywhere.
>
>> which are only used if a GHES platform device is probed. Similarly, the
>> init function also checks for support for firmware first mode.
>> 
>> Create a function, ghes_common_init(), that performs the initialisations
>> and checks that are currently performed on driver init. The function is
>> called when the GHES device is probed.
>> 
>> Delaying initialisation and checks until probe has the added benefit of
>> reducing driver prints on systems that do not support ACPI APEI.
>> 
>> Signed-off-by: Punit Agrawal <punit.agrwal@....com>
>> Cc: Borislav Petkov <bp@...e.de>
>> Cc: James Morse <james.morse@....com>
>> ---
>>  drivers/acpi/apei/ghes.c | 77 +++++++++++++++++++++++++++---------------------
>>  1 file changed, 43 insertions(+), 34 deletions(-)
>> 
>> diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
>> index 007b38abcb34..befb18338acb 100644
>> --- a/drivers/acpi/apei/ghes.c
>> +++ b/drivers/acpi/apei/ghes.c
>> @@ -1088,12 +1088,53 @@ static inline void ghes_nmi_init_cxt(void)
>>  }
>>  #endif /* CONFIG_HAVE_ACPI_APEI_NMI */
>>  
>> +static int ghes_common_init(void)
>> +{
>> +	int rc;
>> +	static bool initialised;
>> +
>> +	if (initialised)
>> +		return 0;
>
> Can't say that I like it. Especially for this "initialised" thing, which
> practically says that this code doesn't conceptually belong here because
> we have to explicitly force it to execute it only once. Even though we
> have execute-once path in ghes_init().

I can see your point. My rationale for the change was that the allocated
resources are only ever useful if there is an actual GHES device in the
system - which we can't know until we hit probe.

In the absence of any GHES entries these are wasted.

>
> What would be much better IMHO is if ghes_init() checked for some APEI
> table which is missing on your system and exited early. That would save
> us all the trouble and solve the situation properly ...

The system does not provide the Hardware Error Source Table (HEST) which
is checked in the hest driver (drivers/acpi/apei/hest.c).

I think I'll go with your original suggestion to change hest_disable
from a boolean to something with more states. Re-checking for the HEST
table again feels like duplication.

Makes sense?

Thanks for taking a look at the patches.

Punit

>
> -- 
> Regards/Gruss,
>     Boris.
>
> SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton, HRB 21284 (AG Nürnberg)