| Message ID | 20160509223924.GX3492@codeaurora.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Hi Stephen, On Mon, May 09, 2016 at 03:39:24PM -0700, Stephen Boyd wrote: > On 05/09, Stephen Boyd wrote: > > > > Ok I applied this one to clk-next. > > > > And I squashed this in to silence the following checker warning. > > drivers/clk/sunxi/clk-sun4i-display.c:110:33: warning: Variable > length array is used. Thanks, just out of curiosity, which checker are you talking about? sparse? Maxime
On Mon, 2016-05-09 at 15:39 -0700, Stephen Boyd wrote: > On 05/09, Stephen Boyd wrote: > > > > > > Ok I applied this one to clk-next. > > > And I squashed this in to silence the following checker warning. > > drivers/clk/sunxi/clk-sun4i-display.c:110:33: warning: Variable > length array is used. > > ---8<--- > diff --git a/drivers/clk/sunxi/clk-sun4i-display.c > b/drivers/clk/sunxi/clk-sun4i-display.c > index f02e250e64ed..f8ff6c4a5633 100644 > --- a/drivers/clk/sunxi/clk-sun4i-display.c > +++ b/drivers/clk/sunxi/clk-sun4i-display.c > @@ -107,7 +107,7 @@ static int sun4i_a10_display_reset_xlate(struct > reset_controller_dev *rcdev, > static void __init sun4i_a10_display_init(struct device_node *node, > const struct > sun4i_a10_display_clk_data *data) > { > - const char *parents[data->parents]; > + const char *parents[4]; > const char *clk_name = node->name; > struct reset_data *reset_data; > struct clk_divider *div = NULL; > @@ -126,8 +126,8 @@ static void __init sun4i_a10_display_init(struct > device_node *node, > return; > } > > - ret = of_clk_parent_fill(node, parents, data->parents); > - if (ret != data->parents) { > + ret = of_clk_parent_fill(node, parents, > ARRAY_SIZE(parents)); > + if (ret != ARRAY_SIZE(parents)) { > pr_err("%s: Could not retrieve the parents\n", > clk_name); > goto unmap; > } This change breaks at least de_[bf]e clocks which have 3 clock parents.
On 05/10, Maxime Ripard wrote: > Hi Stephen, > > On Mon, May 09, 2016 at 03:39:24PM -0700, Stephen Boyd wrote: > > On 05/09, Stephen Boyd wrote: > > > > > > Ok I applied this one to clk-next. > > > > > > > And I squashed this in to silence the following checker warning. > > > > drivers/clk/sunxi/clk-sun4i-display.c:110:33: warning: Variable > > length array is used. > > Thanks, just out of curiosity, which checker are you talking about? > sparse? > I use sparse and smatch. I can't recall which one is complaining here and technically smatch uses sparse underneath.
On 05/10, Priit Laes wrote: > On Mon, 2016-05-09 at 15:39 -0700, Stephen Boyd wrote: > > On 05/09, Stephen Boyd wrote: > > > > > > > > > Ok I applied this one to clk-next. > > > > > And I squashed this in to silence the following checker warning. > > > > drivers/clk/sunxi/clk-sun4i-display.c:110:33: warning: Variable > > length array is used. > > > > ---8<--- > > diff --git a/drivers/clk/sunxi/clk-sun4i-display.c > > b/drivers/clk/sunxi/clk-sun4i-display.c > > index f02e250e64ed..f8ff6c4a5633 100644 > > --- a/drivers/clk/sunxi/clk-sun4i-display.c > > +++ b/drivers/clk/sunxi/clk-sun4i-display.c > > @@ -107,7 +107,7 @@ static int sun4i_a10_display_reset_xlate(struct > > reset_controller_dev *rcdev, > > static void __init sun4i_a10_display_init(struct device_node *node, > > const struct > > sun4i_a10_display_clk_data *data) > > { > > - const char *parents[data->parents]; > > + const char *parents[4]; > > This change breaks at least de_[bf]e clocks which have 3 clock parents. > I just used the largest data->parents number, which was 4. How does that break anything?
On Wed, 2016-05-11 at 15:15 -0700, Stephen Boyd wrote: > On 05/10, Priit Laes wrote: > > > > On Mon, 2016-05-09 at 15:39 -0700, Stephen Boyd wrote: > > > > > > On 05/09, Stephen Boyd wrote: > > > > > > > > > > > > > > > > Ok I applied this one to clk-next. > > > > > > > And I squashed this in to silence the following checker warning. > > > > > > drivers/clk/sunxi/clk-sun4i-display.c:110:33: warning: Variable > > > length array is used. > > > > > > ---8<--- > > > diff --git a/drivers/clk/sunxi/clk-sun4i-display.c > > > b/drivers/clk/sunxi/clk-sun4i-display.c > > > index f02e250e64ed..f8ff6c4a5633 100644 > > > --- a/drivers/clk/sunxi/clk-sun4i-display.c > > > +++ b/drivers/clk/sunxi/clk-sun4i-display.c > > > @@ -107,7 +107,7 @@ static int > > > sun4i_a10_display_reset_xlate(struct > > > reset_controller_dev *rcdev, > > > static void __init sun4i_a10_display_init(struct device_node > > > *node, > > > const struct > > > sun4i_a10_display_clk_data *data) > > > { > > > - const char *parents[data->parents]; > > > + const char *parents[4]; > > This change breaks at least de_[bf]e clocks which have 3 clock > > parents. > I just used the largest data->parents number, which was 4. How > does that break anything? If you look at the sun4i_a10_display_init, it contains this block: ret = of_clk_parent_fill(node, parents, ARRAY_SIZE(parents)); if (ret != ARRAY_SIZE(parents)) { pr_err("%s: Could not retrieve the parents\n", clk_name); goto unmap; } of_clk_parent_fill returns 3 for de_be/de_fe nodes, and ARRAY_SIZE(parents) is 4. Päikest, Priit :)
On Thu, May 12, 2016 at 06:39:20AM +0300, Priit Laes wrote: > On Wed, 2016-05-11 at 15:15 -0700, Stephen Boyd wrote: > > On 05/10, Priit Laes wrote: > > > > > > On Mon, 2016-05-09 at 15:39 -0700, Stephen Boyd wrote: > > > > > > > > On 05/09, Stephen Boyd wrote: > > > > > > > > > > > > > > > > > > > > Ok I applied this one to clk-next. > > > > > > > > > And I squashed this in to silence the following checker warning. > > > > > > > > drivers/clk/sunxi/clk-sun4i-display.c:110:33: warning: Variable > > > > length array is used. > > > > > > > > ---8<--- > > > > diff --git a/drivers/clk/sunxi/clk-sun4i-display.c > > > > b/drivers/clk/sunxi/clk-sun4i-display.c > > > > index f02e250e64ed..f8ff6c4a5633 100644 > > > > --- a/drivers/clk/sunxi/clk-sun4i-display.c > > > > +++ b/drivers/clk/sunxi/clk-sun4i-display.c > > > > @@ -107,7 +107,7 @@ static int > > > > sun4i_a10_display_reset_xlate(struct > > > > reset_controller_dev *rcdev, > > > > static void __init sun4i_a10_display_init(struct device_node > > > > *node, > > > > const struct > > > > sun4i_a10_display_clk_data *data) > > > > { > > > > - const char *parents[data->parents]; > > > > + const char *parents[4]; > > > This change breaks at least de_[bf]e clocks which have 3 clock > > > parents. > > > I just used the largest data->parents number, which was 4. How > > does that break anything? > > If you look at the sun4i_a10_display_init, it contains this block: > > ret = of_clk_parent_fill(node, parents, ARRAY_SIZE(parents)); > if (ret != ARRAY_SIZE(parents)) { > pr_err("%s: Could not retrieve the parents\n", clk_name); > goto unmap; > } > > of_clk_parent_fill returns 3 for de_be/de_fe nodes, and > ARRAY_SIZE(parents) is 4. Replacing both ARRAY_SIZE(parents) by data->parents would work though. Maxime
diff --git a/drivers/clk/sunxi/clk-sun4i-display.c b/drivers/clk/sunxi/clk-sun4i-display.c index f02e250e64ed..f8ff6c4a5633 100644 --- a/drivers/clk/sunxi/clk-sun4i-display.c +++ b/drivers/clk/sunxi/clk-sun4i-display.c @@ -107,7 +107,7 @@ static int sun4i_a10_display_reset_xlate(struct reset_controller_dev *rcdev, static void __init sun4i_a10_display_init(struct device_node *node, const struct sun4i_a10_display_clk_data *data) { - const char *parents[data->parents]; + const char *parents[4]; const char *clk_name = node->name; struct reset_data *reset_data; struct clk_divider *div = NULL; @@ -126,8 +126,8 @@ static void __init sun4i_a10_display_init(struct device_node *node, return; } - ret = of_clk_parent_fill(node, parents, data->parents); - if (ret != data->parents) { + ret = of_clk_parent_fill(node, parents, ARRAY_SIZE(parents)); + if (ret != ARRAY_SIZE(parents)) { pr_err("%s: Could not retrieve the parents\n", clk_name); goto unmap; }