linuxppc-dev.lists.ozlabs.org archive mirror
 help / color / mirror / Atom feed
From: Christian Zigotzky <chzigotzky@xenosoft.de>
To: Benjamin Herrenschmidt <benh@kernel.crashing.org>,
	Scott Wood <oss@buserror.net>,
	LEROY Christophe <christophe.leroy@c-s.fr>
Cc: linuxppc-dev@lists.ozlabs.org
Subject: Re: NXP P50XX/e5500: SMP doesn't work anymore with the latest Git kernel
Date: Tue, 30 Oct 2018 16:05:19 +0100	[thread overview]
Message-ID: <4fceca56-f4c9-1bda-7504-3c82f862a522@xenosoft.de> (raw)
In-Reply-To: <95aab857-7bc5-8585-e426-7a0e60792934@xenosoft.de>

On 30 October 2018 at 08:27AM, Christian Zigotzky wrote:
> On 30 October 2018 at 02:59AM, Benjamin Herrenschmidt wrote:
>> On Tue, 2018-10-30 at 02:42 +0100, Christian Zigotzky wrote:
>>> OF patch for the latest Git kernel: http://www.xenosoft.de/of_v2.patch
>> This just seems to revert a whole bunch of stuff, not really the right
>> way to go. Why is of_get_next_cpu_node() not finding your CPUs ? There
>> must be something wrong with the device-tree...
> Maybe we need a new dtb for QEMU ppce500 and P50XX machines. I don't 
> know. For us it is good to have this patch because we can test 
> further. (other updates)
> With this patch, SMP works again on virtual QEMU ppce500 and P50XX 
> machines. If you have another code, then I will test it with QEMU and 
> my P5020 board.
>
> Thanks,
> Christian
We use the 'cyrus_p5020.dts' file for creating the dtb file for our 
P5020 board. You can find it in the official kernel source code. 
(arch/powerpc/boot/dts/cyrus_p5020.dts)
For the virtual ppce500 machine I don't need to add a dtb file to the 
QEMU command. But maybe there is a solution:

On 30 October 2018 at 3:18PM, Michael Ellerman wrote:
> Hi Rob,
>
> Sorry I missed this when you posted it.
>
> Rob Herring <robh@kernel.org> writes:
>> Iterating thru cpu nodes is a common pattern. Create a common iterator
>> which can find child nodes either by node name or device_type == cpu.
>> Using the former will allow for eventually dropping device_type
>> properties which are deprecated for FDT.
> Device trees we see on powerpc generally don't (never?) use "cpu" as the
> node name for CPU nodes. And many of those device trees come from
> firmware, so we can't update them.
>
> So dropping support for device_type is a non-starter from our POV.
>
> cheers
>
>
>>
>>> ----------------------------- of_v2.patch -----------------------------
>>>
>>> diff -rupN a/drivers/of/base.c b/drivers/of/base.c
>>> --- a/drivers/of/base.c    2018-10-30 02:19:30.827089495 +0100
>>> +++ b/drivers/of/base.c    2018-10-30 02:18:51.666856715 +0100
>>> @@ -395,7 +395,7 @@ struct device_node *of_get_cpu_node(int
>>>    {
>>>        struct device_node *cpun;
>>>
>>> -    for_each_of_cpu_node(cpun) {
>>> +    for_each_node_by_type(cpun, "cpu") {
>>>            if (arch_find_n_match_cpu_physical_id(cpun, cpu, thread))
>>>                return cpun;
>>>        }
>>> @@ -750,45 +750,6 @@ struct device_node *of_get_next_availabl
>>>    EXPORT_SYMBOL(of_get_next_available_child);
>>>
>>>    /**
>>> - *    of_get_next_cpu_node - Iterate on cpu nodes
>>> - *    @prev:    previous child of the /cpus node, or NULL to get first
>>> - *
>>> - *    Returns a cpu node pointer with refcount incremented, use
>>> of_node_put()
>>> - *    on it when done. Returns NULL when prev is the last child. 
>>> Decrements
>>> - *    the refcount of prev.
>>> - */
>>> -struct device_node *of_get_next_cpu_node(struct device_node *prev)
>>> -{
>>> -    struct device_node *next = NULL;
>>> -    unsigned long flags;
>>> -    struct device_node *node;
>>> -
>>> -    if (!prev)
>>> -        node = of_find_node_by_path("/cpus");
>>> -
>>> -    raw_spin_lock_irqsave(&devtree_lock, flags);
>>> -    if (prev)
>>> -        next = prev->sibling;
>>> -    else if (node) {
>>> -        next = node->child;
>>> -        of_node_put(node);
>>> -    }
>>> -    for (; next; next = next->sibling) {
>>> -        if (!(of_node_name_eq(next, "cpu") ||
>>> -              (next->type && !of_node_cmp(next->type, "cpu"))))
>>> -            continue;
>>> -        if (!__of_device_is_available(next))
>>> -            continue;
>>> -        if (of_node_get(next))
>>> -            break;
>>> -    }
>>> -    of_node_put(prev);
>>> -    raw_spin_unlock_irqrestore(&devtree_lock, flags);
>>> -    return next;
>>> -}
>>> -EXPORT_SYMBOL(of_get_next_cpu_node);
>>> -
>>> -/**
>>>     * of_get_compatible_child - Find compatible child node
>>>     * @parent:    parent node
>>>     * @compatible:    compatible string
>>> diff -rupN a/include/linux/of.h b/include/linux/of.h
>>> --- a/include/linux/of.h    2018-10-30 02:19:32.047096634 +0100
>>> +++ b/include/linux/of.h    2018-10-30 02:18:51.666856715 +0100
>>> @@ -347,7 +347,6 @@ extern const void *of_get_property(const
>>>                    const char *name,
>>>                    int *lenp);
>>>    extern struct device_node *of_get_cpu_node(int cpu, unsigned int 
>>> *thread);
>>> -extern struct device_node *of_get_next_cpu_node(struct device_node 
>>> *prev);
>>>
>>>    #define for_each_property_of_node(dn, pp) \
>>>        for (pp = dn->properties; pp != NULL; pp = pp->next)
>>> @@ -757,11 +756,6 @@ static inline struct device_node *of_get
>>>        return NULL;
>>>    }
>>>
>>> -static inline struct device_node *of_get_next_cpu_node(struct
>>> device_node *prev)
>>> -{
>>> -    return NULL;
>>> -}
>>> -
>>>    static inline int of_n_addr_cells(struct device_node *np)
>>>    {
>>>        return 0;
>>> @@ -1239,10 +1233,6 @@ static inline int of_property_read_s32(c
>>>        for (child = of_get_next_available_child(parent, NULL); child !=
>>> NULL; \
>>>             child = of_get_next_available_child(parent, child))
>>>
>>> -#define for_each_of_cpu_node(cpu) \
>>> -    for (cpu = of_get_next_cpu_node(NULL); cpu != NULL; \
>>> -         cpu = of_get_next_cpu_node(cpu))
>>> -
>>>    #define for_each_node_with_property(dn, prop_name) \
>>>        for (dn = of_find_node_with_property(NULL, prop_name); dn; \
>>>             dn = of_find_node_with_property(dn, prop_name))
>>
>
>


  reply	other threads:[~2018-10-30 15:09 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-10-28 16:35 NXP P50XX/e5500: SMP doesn't work anymore with the latest Git kernel Christian Zigotzky
2018-10-29  9:56 ` Christian Zigotzky
2018-10-29 10:22   ` arch/powerpc/kvm/trace.h:9:0: error: "TRACE_INCLUDE_PATH" redefined Christian Zigotzky
2018-10-31 10:09     ` Christian Zigotzky
2018-11-05  7:09       ` Christian Zigotzky
2018-10-29 17:00   ` NXP P50XX/e5500: SMP doesn't work anymore with the latest Git kernel Christian Zigotzky
2018-10-29 18:13     ` LEROY Christophe
2018-10-29 20:12       ` Christian Zigotzky
2018-10-29 23:04         ` Scott Wood
2018-10-30  0:41           ` Christian Zigotzky
2018-10-30  1:42             ` Christian Zigotzky
2018-10-30  1:59               ` Benjamin Herrenschmidt
2018-10-30  7:27                 ` Christian Zigotzky
2018-10-30 15:05                   ` Christian Zigotzky [this message]
2018-10-29 22:24     ` Christian Zigotzky
2018-10-31 13:20       ` Michael Ellerman
2018-10-31 13:38         ` Christian Zigotzky
2018-11-07  2:11 ` Scott Wood
2018-11-07  6:46   ` Christian Zigotzky

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4fceca56-f4c9-1bda-7504-3c82f862a522@xenosoft.de \
    --to=chzigotzky@xenosoft.de \
    --cc=benh@kernel.crashing.org \
    --cc=christophe.leroy@c-s.fr \
    --cc=linuxppc-dev@lists.ozlabs.org \
    --cc=oss@buserror.net \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).