| Message ID | 20220616044915.3978097-1-windhl@126.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v2] soc: sifive: (sifive_l2_cache) Add missing of_node_put() | expand |
At 2022-06-16 12:49:15, "Liang He" <windhl@126.com> wrote: >In sifive_l2_init(), of_find_matching_node() will return a node pointer >with refcount incremented. We should use of_node_put() in each fail path >or when it is not used anymore. > >Reported-by: kernel test robot <lkp@intel.com> > >Signed-off-by: Liang He <windhl@126.com> >--- Sorry, please ignore this version as it still has a bug. I will have a strict check and compile the code before my next commit.
On 16/06/2022 06:54, Liang He wrote: > EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe > > At 2022-06-16 12:49:15, "Liang He" <windhl@126.com> wrote: >> In sifive_l2_init(), of_find_matching_node() will return a node pointer >> with refcount incremented. We should use of_node_put() in each fail path >> or when it is not used anymore. >> >> Reported-by: kernel test robot <lkp@intel.com> >> >> Signed-off-by: Liang He <windhl@126.com> >> --- > > > Sorry, please ignore this version as it still has a bug. I will have a strict check and compile the code before my next commit. While you're at it - drop the Reported-by too. You only need to include that if LKP found a problem with a patch that's been applied already. If it's on something in progress, just fix what it pointed out & resend. Thanks, Conor.
On 16/06/2022 05:49, Liang He wrote: > [PATCH v2] soc: sifive: (sifive_l2_cache) Add missing of_node_put() I figure you submitted a patch against hwmon & Guenter asked for you to change the subject line? This is the hwmon style & the brackets aren't needed in (most) other subsystems. Thanks, Conor. > [You don't often get email from windhl@126.com. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ] > > EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe > > In sifive_l2_init(), of_find_matching_node() will return a node pointer > with refcount incremented. We should use of_node_put() in each fail path > or when it is not used anymore. > > Reported-by: kernel test robot <lkp@intel.com> > > Signed-off-by: Liang He <windhl@126.com> > --- > changelog: > > v2: (1) fix bug, introduced by v1 patch, reporeted by kernel-test-robot > v1: fix missing bug > > ps: Because of many local commit, when using --base for format-patch, > there are too many prerequest-patch-id. I wonder if it will lead to other > problems. So now, I send this commit still without --base. > > drivers/soc/sifive/sifive_l2_cache.c | 24 +++++++++++++++++------- > 1 file changed, 17 insertions(+), 7 deletions(-) > > diff --git a/drivers/soc/sifive/sifive_l2_cache.c b/drivers/soc/sifive/sifive_l2_cache.c > index 59640a1d0b28..e0e3d8b64c25 100644 > --- a/drivers/soc/sifive/sifive_l2_cache.c > +++ b/drivers/soc/sifive/sifive_l2_cache.c > @@ -202,17 +202,22 @@ static int __init sifive_l2_init(void) > if (!np) > return -ENODEV; > > - if (of_address_to_resource(np, 0, &res)) > - return -ENODEV; > + if (of_address_to_resource(np, 0, &res)) { > + ret = -ENODEV; > + goto out_put; > + } > > l2_base = ioremap(res.start, resource_size(&res)); > - if (!l2_base) > - return -ENOMEM; > + if (!l2_base) { > + ret = -ENOMEM; > + goto out_put; > + } > > intr_num = of_property_count_u32_elems(np, "interrupts"); > if (!intr_num) { > pr_err("L2CACHE: no interrupts property\n"); > - return -ENODEV; > + ret = -ENODEV; > + goto out_put; > } > > for (i = 0; i < intr_num; i++) { > @@ -220,7 +225,8 @@ static int __init sifive_l2_init(void) > rc = request_irq(g_irq[i], l2_int_handler, 0, "l2_ecc", NULL); > if (rc) { > pr_err("L2CACHE: Could not request IRQ %d\n", g_irq[i]); > - return rc; > + ret = rc; > + goto out_put; > } > } > > @@ -232,6 +238,10 @@ static int __init sifive_l2_init(void) > #ifdef CONFIG_DEBUG_FS > setup_sifive_debug(); > #endif > - return 0; > + ret = 0; > + > +out_put: > + of_node_put(np); > + return ret; > } > device_initcall(sifive_l2_init); > -- > 2.25.1 > > > _______________________________________________ > linux-riscv mailing list > linux-riscv@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/linux-riscv
At 2022-06-16 14:26:17, Conor.Dooley@microchip.com wrote: >On 16/06/2022 06:54, Liang He wrote: >> EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe >> >> At 2022-06-16 12:49:15, "Liang He" <windhl@126.com> wrote: >>> In sifive_l2_init(), of_find_matching_node() will return a node pointer >>> with refcount incremented. We should use of_node_put() in each fail path >>> or when it is not used anymore. >>> >>> Reported-by: kernel test robot <lkp@intel.com> >>> >>> Signed-off-by: Liang He <windhl@126.com> >>> --- >> >> >> Sorry, please ignore this version as it still has a bug. I will have a strict check and compile the code before my next commit. > >While you're at it - drop the Reported-by too. >You only need to include that if LKP found a problem with >a patch that's been applied already. If it's on something >in progress, just fix what it pointed out & resend. > >Thanks, >Conor. > Thanks for all your help, Conor. I will make a proper patch considering all things you suggested.
diff --git a/drivers/soc/sifive/sifive_l2_cache.c b/drivers/soc/sifive/sifive_l2_cache.c index 59640a1d0b28..e0e3d8b64c25 100644 --- a/drivers/soc/sifive/sifive_l2_cache.c +++ b/drivers/soc/sifive/sifive_l2_cache.c @@ -202,17 +202,22 @@ static int __init sifive_l2_init(void) if (!np) return -ENODEV; - if (of_address_to_resource(np, 0, &res)) - return -ENODEV; + if (of_address_to_resource(np, 0, &res)) { + ret = -ENODEV; + goto out_put; + } l2_base = ioremap(res.start, resource_size(&res)); - if (!l2_base) - return -ENOMEM; + if (!l2_base) { + ret = -ENOMEM; + goto out_put; + } intr_num = of_property_count_u32_elems(np, "interrupts"); if (!intr_num) { pr_err("L2CACHE: no interrupts property\n"); - return -ENODEV; + ret = -ENODEV; + goto out_put; } for (i = 0; i < intr_num; i++) { @@ -220,7 +225,8 @@ static int __init sifive_l2_init(void) rc = request_irq(g_irq[i], l2_int_handler, 0, "l2_ecc", NULL); if (rc) { pr_err("L2CACHE: Could not request IRQ %d\n", g_irq[i]); - return rc; + ret = rc; + goto out_put; } } @@ -232,6 +238,10 @@ static int __init sifive_l2_init(void) #ifdef CONFIG_DEBUG_FS setup_sifive_debug(); #endif - return 0; + ret = 0; + +out_put: + of_node_put(np); + return ret; } device_initcall(sifive_l2_init);
In sifive_l2_init(), of_find_matching_node() will return a node pointer with refcount incremented. We should use of_node_put() in each fail path or when it is not used anymore. Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Liang He <windhl@126.com> --- changelog: v2: (1) fix bug, introduced by v1 patch, reporeted by kernel-test-robot v1: fix missing bug ps: Because of many local commit, when using --base for format-patch, there are too many prerequest-patch-id. I wonder if it will lead to other problems. So now, I send this commit still without --base. drivers/soc/sifive/sifive_l2_cache.c | 24 +++++++++++++++++------- 1 file changed, 17 insertions(+), 7 deletions(-)