[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <56B88C43.2070904@redhat.com>
Date: Mon, 8 Feb 2016 13:38:27 +0100
From: Tomas Henzl <thenzl@...hat.com>
To: Suganath Prabu Subramani <suganath-prabu.subramani@...adcom.com>
Cc: Suganath prabu Subaramani
<suganath-prabu.subramani@...gotech.com>, JBottomley@...allels.com,
jejb@...nel.org, Christoph Hellwig <hch@...radead.org>,
martin.petersen@...cle.com, linux-scsi@...r.kernel.org,
Sathya Prakash <Sathya.Prakash@...gotech.com>,
Kashyap Desai <kashyap.desai@...gotech.com>,
Krishnaraddi Mankani <krishnaraddi.mankani@...gotech.com>,
linux-kernel@...r.kernel.org,
Chaitra Basappa <chaitra.basappa@...gotech.com>,
Sreekanth Reddy <sreekanth.reddy@...gotech.com>,
"Elliott, Robert (Server Storage)" <elliott@...com>
Subject: Re: [mpt3sas driver 06/10] mpt3sas: Added smp_affinity_enable module
parameter.
On 8.2.2016 07:01, Suganath Prabu Subramani wrote:
> Hi Tomas,
> Initially, we posted this patch with "zalloc_cpumassk_var" and Robert
> has suggested to use "alloc_cpumask_var". Please check the below link.
> "http://www.gossamer-threads.com/lists/linux/kernel/2068280?do=post_view_threaded#2068280"
That thread is about allocating more than KMALLOC_MAX_SIZE with a kmalloc,
it doesn't matter if it is a kzalloc or kmalloc.
If that is an issue in mpt3sas, in your code fails already here when !CONFIG_CPUMASK_OFFSTACK
reply_q = kzalloc(sizeof(struct adapter_reply_queue), GFP_KERNEL);
....
Good news is that the the size allocated in your case is much smaller than what
the lpfc tried at that time, (it used a struct with struct cpumask maskbits;
and multiplied it with nr_cpus) -> this is not your case.
You can use that zalloc_cpumask_var.
--tm
>
> We incorporated other review comments for freeing reply_q and
> cpumask_var in this patch and sending it soon.
>
> Thanks,
> Suganath prabu S
>
> On Thu, Feb 4, 2016 at 8:31 PM, Tomas Henzl <thenzl@...hat.com> wrote:
>> On 28.1.2016 07:37, Suganath prabu Subaramani wrote:
>>> From: Suganath prabu Subramani <suganath-prabu.subramani@...gotech.com>
>>>
>>> Module parameter to enable/disable configuring
>>> affinity hint for msix vector.
>>> SMP affinity feature can be enabled/disabled by setting
>>> module parameter "smp_affinity_enable" to 1/0.
>>> By default this feature is enabled. (smp_affinity_enable = 1 enabled).
>>>
>>> Signed-off-by: Suganath prabu Subramani <suganath-prabu.subramani@...gotech.com>
>>> Signed-off-by: Chaitra P B <chaitra.basappa@...gotech.com>
>>> ---
>>> drivers/scsi/mpt3sas/mpt3sas_base.c | 34 +++++++++++++++++++++++-----------
>>> 1 file changed, 23 insertions(+), 11 deletions(-)
>>>
>>> diff --git a/drivers/scsi/mpt3sas/mpt3sas_base.c b/drivers/scsi/mpt3sas/mpt3sas_base.c
>>> index 31838d9a..a1a3b39 100644
>>> --- a/drivers/scsi/mpt3sas/mpt3sas_base.c
>>> +++ b/drivers/scsi/mpt3sas/mpt3sas_base.c
>>> @@ -83,6 +83,10 @@ static int msix_disable = -1;
>>> module_param(msix_disable, int, 0);
>>> MODULE_PARM_DESC(msix_disable, " disable msix routed interrupts (default=0)");
>>>
>>> +static int smp_affinity_enable = 1;
>>> +module_param(smp_affinity_enable, int, S_IRUGO);
>>> +MODULE_PARM_DESC(smp_affinity_enable, "SMP affinity feature enable/disbale Default: enable(1)");
>>> +
>>> static int max_msix_vectors = -1;
>>> module_param(max_msix_vectors, int, 0);
>>> MODULE_PARM_DESC(max_msix_vectors,
>>> @@ -1812,8 +1816,10 @@ _base_free_irq(struct MPT3SAS_ADAPTER *ioc)
>>>
>>> list_for_each_entry_safe(reply_q, next, &ioc->reply_queue_list, list) {
>>> list_del(&reply_q->list);
>>> - irq_set_affinity_hint(reply_q->vector, NULL);
>>> - free_cpumask_var(reply_q->affinity_hint);
>>> + if (smp_affinity_enable) {
>>> + irq_set_affinity_hint(reply_q->vector, NULL);
>>> + free_cpumask_var(reply_q->affinity_hint);
>>> + }
>>> synchronize_irq(reply_q->vector);
>>> free_irq(reply_q->vector, reply_q);
>>> kfree(reply_q);
>>> @@ -1844,9 +1850,11 @@ _base_request_irq(struct MPT3SAS_ADAPTER *ioc, u8 index, u32 vector)
>>> reply_q->msix_index = index;
>>> reply_q->vector = vector;
>>>
>>> - if (!alloc_cpumask_var(&reply_q->affinity_hint, GFP_KERNEL))
>>> - return -ENOMEM;
>>> - cpumask_clear(reply_q->affinity_hint);
>>> + if (smp_affinity_enable) {
>>> + if (!alloc_cpumask_var(&reply_q->affinity_hint, GFP_KERNEL))
>> a kfree(reply_q); should go here
>>
>>> + return -ENOMEM;
>>> + cpumask_clear(reply_q->affinity_hint);
>> maybe zalloc_cpumask_var ?
>>
>>> + }
>> if (r) {
>> pr_err(MPT3SAS_FMT "unable to allocate interrupt %d!\n",
>> reply_q->name, vector);
>> kfree(reply_q);
>> and a free_cpumask_var (..) here ?
>> return -EBUSY;
>>
>>
>> cheers,
>> tomash
>>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> the body of a message to majordomo@...r.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists