linux-kernel - Re: [PATCH v11 2/4] clk: Make clk API return per-user struct clk instances

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]

Message-ID: <54C0DBD6.2000502@collabora.com>
Date:	Thu, 22 Jan 2015 12:15:34 +0100
From:	Tomeu Vizoso <tomeu.vizoso@...labora.com>
To:	Stephen Boyd <sboyd@...eaurora.org>
CC:	linux-kernel@...r.kernel.org,
	Mike Turquette <mturquette@...aro.org>,
	Javier Martinez Canillas <javier.martinez@...labora.co.uk>,
	Paul Walmsley <paul@...an.com>,
	Tony Lindgren <tony@...mide.com>,
	Russell King <linux@....linux.org.uk>,
	linux-omap@...r.kernel.org, linux-arm-kernel@...ts.infradead.org
Subject: Re: [PATCH v11 2/4] clk: Make clk API return per-user struct clk
 instances

On 01/22/2015 02:01 AM, Stephen Boyd wrote:
> On 01/21, Tomeu Vizoso wrote:
>> @@ -2075,10 +2210,12 @@ struct clk *clk_register(struct device *dev, struct clk_hw *hw)
>>  		}
>>  	}
>>  
>> -	ret = __clk_init(dev, clk);
>> +	hw->clk = __clk_create_clk(hw, NULL, NULL);
>> +	ret = __clk_init(dev, hw->clk);
>>  	if (!ret)
>> -		return clk;
>> +		return hw->clk;
>>  
>> +	kfree(hw->clk);
>>  fail_parent_names_copy:
>>  	while (--i >= 0)
>>  		kfree(clk->parent_names[i]);
> 
> Sigh, this patch is so huge I keep finding more things. Sorry. It
> looks like __clk_create_clk() can return an error pointer, which
> we then send directly to __clk_init. First off, we shouldn't
> kfree() that pointer if it's an error pointer. Second, we
> shouldn't crash in __clk_init() in such a situation so there
> needs to be some sort of check somewhere.

Oops, done. I have reused the fail_parent_names_copy label as the
less-bad possibility. Probably the error labels should be named after
the target code and not after what the source code does, as per the
latest CodingStyle additions.

> BTW, please try and fixup checkpatch warnings.

What were you thinking of specifically? I'm running it with
--max-line-length=106 and the other warnings are in clk-test.c that I
still have to polish when I get some time.

>> diff --git a/drivers/clk/clkdev.c b/drivers/clk/clkdev.c
>> index da4bda8..fac3244 100644
>> --- a/drivers/clk/clkdev.c
>> +++ b/drivers/clk/clkdev.c
>> @@ -69,20 +70,22 @@ struct clk *of_clk_get(struct device_node *np, int index)
> [...]
>> -struct clk *of_clk_get_by_name(struct device_node *np, const char *name)
>> +static struct clk *__of_clk_get_by_name(struct device_node *np, const char *name)
> 
> It would be nice if this returned an already __clk_create_clk()ed
> pointer.
> 
>>  {
>>  	struct clk *clk = ERR_PTR(-ENOENT);
>>  
>> @@ -119,7 +122,33 @@ struct clk *of_clk_get_by_name(struct device_node *np, const char *name)
> [...]
>> +struct clk *of_clk_get_by_name(struct device_node *np, const char *name)
>> +{
>> +	struct clk *clk = __of_clk_get_by_name(np, name);
>> +
>> +	if (!IS_ERR(clk))
>> +		clk = __clk_create_clk(__clk_get_hw(clk), np->full_name, name);
> 
> Because we do it here where we know we're CONFIG_COMMON_CLK=y.
> 
>> +
>> +	return clk;
>> +}
>>  EXPORT_SYMBOL(of_clk_get_by_name);
>> +
>> +#else /* defined(CONFIG_OF) && defined(CONFIG_COMMON_CLK) */
>> +
>> +static struct clk *__of_clk_get_by_name(struct device_node *np, const char *name)
>> +{
>> +	return ERR_PTR(-ENOENT);
>> +}
>>  #endif
>>  
>>  /*
>> @@ -185,9 +229,13 @@ struct clk *clk_get(struct device *dev, const char *con_id)
>>  	struct clk *clk;
>>  
>>  	if (dev) {
>> -		clk = of_clk_get_by_name(dev->of_node, con_id);
>> -		if (!IS_ERR(clk))
>> +		clk = __of_clk_get_by_name(dev->of_node, con_id);
>> +		if (!IS_ERR(clk)) {
>> +#if defined(CONFIG_COMMON_CLK)
>> +			clk = __clk_create_clk(__clk_get_hw(clk), dev_id, con_id);
>> +#endif
> 
> And we do it here where we could remove the #ifdef.

Yeah, I tried to reduce the ifdefing back then and this is the simplest
I could come up with. The reason for clk_get() to call
__clk_create_clk() directly is that it has more relevant information
with which to tag the per-user clk.

of_clk_get_by_name() has the name of the node but not the dev_id, which
in my testing looked as much less useful when debugging who did what to
a clock.

Thanks,

Tomeu

>>  			return clk;
>> +		}
>>  		if (PTR_ERR(clk) == -EPROBE_DEFER)
>>  			return clk;
>>  	}
> 

--
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/