From: "Liang He" <windhl@126.com>
To: "Orson Zhai" <orsonzhai@gmail.com>
Cc: mturquette@baylibre.com, "Stephen Boyd" <sboyd@kernel.org>,
"Baolin Wang" <baolin.wang7@gmail.com>,
"Lyra Zhang" <zhang.lyra@gmail.com>,
linux-clk@vger.kernel.org
Subject: Re:Re: Re: [PATCH] clk/sprd: Hold reference returned by of_get_parent()
Date: Sun, 26 Jun 2022 12:03:31 +0800 (CST) [thread overview]
Message-ID: <1a1ed744.abe.1819e2dd0bc.Coremail.windhl@126.com> (raw)
In-Reply-To: <CA+H2tpHyXD1pr7G9nxU9tJPgtfE5MioFE+Dy+bUwz+z=ap-MXg@mail.gmail.com>
At 2022-06-26 11:32:40, "Orson Zhai" <orsonzhai@gmail.com> wrote:
>Hi Liang,
>
>On Sat, Jun 25, 2022 at 11:34 PM Liang He <windhl@126.com> wrote:
>>
>>
>> At 2022-06-25 16:36:04, "Orson Zhai" <orsonzhai@gmail.com> wrote:
>> >Hi Liang,
>> >
>> >On Fri, Jun 24, 2022 at 6:39 PM Liang He <windhl@126.com> wrote:
>> >>
>> >> We should hold the reference returned by of_get_parent() and use it
>> >> to call of_node_out() for refcount balance.
>> >>
>> >> Fixes: f95e8c7923d1 ("clk: sprd: support to get regmap from parent node")
>> >> Signed-off-by: Liang He <windhl@126.com>
>> >> ---
>> >> these bugs are found in 5.19rc2
>> >> these bugs are confirmed not fixed by lore.kernel.org
>> >> these bugs are compiled tested in 5.19rc2 with at91_dt_defconfig
>> >>
>> >> drivers/clk/sprd/common.c | 16 +++++++++++-----
>> >> 1 file changed, 11 insertions(+), 5 deletions(-)
>> >>
>> >> diff --git a/drivers/clk/sprd/common.c b/drivers/clk/sprd/common.c
>> >> index d620bbbcdfc8..08c1d7a9ec8b 100644
>> >> --- a/drivers/clk/sprd/common.c
>> >> +++ b/drivers/clk/sprd/common.c
>> >> @@ -41,35 +41,41 @@ int sprd_clk_regmap_init(struct platform_device *pdev,
>> >> {
>> >> void __iomem *base;
>> >> struct device *dev = &pdev->dev;
>> >> - struct device_node *node = dev->of_node;
>> >> + struct device_node *node = dev->of_node, *np;
>> >> struct regmap *regmap;
>> >>
>> >> + np = of_get_parent(dev->of_node);
>> >> if (of_find_property(node, "sprd,syscon", NULL)) {
>> >> regmap = syscon_regmap_lookup_by_phandle(node, "sprd,syscon");
>> >
>> >Why not to call of_node_put() in this function?
>> >He returns an error and it is his responsibility to clean up
>> >everything he used, right?
>> >
>> >-Orson
>> >
>> Hi, Orson,
>>
>> I only know we should use the pair of functions (of_get_parent(), of_node_put)
>> to keep refcount balance.
>>
>> However, as I do not really know the semantic of other functions, so I cannot
>> give other kinds of patch.
>
>Sorry, I didn't catch your point before.
>You're right on this unbalance. But it has no direct relation to the
>function here.
Thanks for your understanding. And yes, it has no direct relation
to the function here.
>I think it is better to minimize the effective range of of_get_parent()
>
>Could you refer to Pantelis's answer to the last question at [1].
>
>Thanks.
>-Orson
>
>[1] https://www.spinics.net/lists/devicetree/msg122555.html
>
Thanks, Orson.
Yes, I also wanted to make a better patch solution as the bug is only caused
by the 'else if' branch.
However, I cannot find a better solution to fix the bug in the middle of nested
'if-else if-else' brances unless we try to rewrite them, which is more difficult for me, now.
But if you think it is (semantic) safe to rewrite these branches, I can try it.
Thanks again for your relpy.
Liang
>>
>> By the way, if we chose to put of_node_put() in this function, I think it also need
>> to put of_node_put() into 'device_node_to_regmap' as it also return error code in
>> following path.
>>
>> Anyway, thanks very much for your reply and review on my patch code.
>>
>> Liang
>>
>>
>> >> if (IS_ERR(regmap)) {
>> >> pr_err("%s: failed to get syscon regmap\n", __func__);
>> >> + of_node_put(np);
>> >> return PTR_ERR(regmap);
>> >> }
>> >> - } else if (of_device_is_compatible(of_get_parent(dev->of_node),
>> >> - "syscon")) {
>> >> - regmap = device_node_to_regmap(of_get_parent(dev->of_node));
>> >> + } else if (of_device_is_compatible(np, "syscon")) {
>> >> + regmap = device_node_to_regmap(np);
>> >> if (IS_ERR(regmap)) {
>> >> + of_node_put(np);
>> >> dev_err(dev, "failed to get regmap from its parent.\n");
>> >> return PTR_ERR(regmap);
>> >> }
>> >> } else {
>> >> base = devm_platform_ioremap_resource(pdev, 0);
>> >> - if (IS_ERR(base))
>> >> + if (IS_ERR(base)) {
>> >> + of_node_put(np);
>> >> return PTR_ERR(base);
>> >> + }
>> >>
>> >> regmap = devm_regmap_init_mmio(&pdev->dev, base,
>> >> &sprdclk_regmap_config);
>> >> if (IS_ERR(regmap)) {
>> >> + of_node_put(np);
>> >> pr_err("failed to init regmap\n");
>> >> return PTR_ERR(regmap);
>> >> }
>> >> }
>> >>
>> >> + of_node_put(np);
>> >> sprd_clk_set_regmap(desc, regmap);
>> >>
>> >> return 0;
>> >> --
>> >> 2.25.1
>> >>
next prev parent reply other threads:[~2022-06-26 4:03 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-06-24 10:38 [PATCH] clk/sprd: Hold reference returned by of_get_parent() Liang He
2022-06-25 8:36 ` Orson Zhai
2022-06-25 15:32 ` Liang He
2022-06-26 3:32 ` Orson Zhai
2022-06-26 4:03 ` Liang He [this message]
2022-06-27 7:55 ` Orson Zhai
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=1a1ed744.abe.1819e2dd0bc.Coremail.windhl@126.com \
--to=windhl@126.com \
--cc=baolin.wang7@gmail.com \
--cc=linux-clk@vger.kernel.org \
--cc=mturquette@baylibre.com \
--cc=orsonzhai@gmail.com \
--cc=sboyd@kernel.org \
--cc=zhang.lyra@gmail.com \
/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