| Message ID | 20201217204236.163446-1-christophe.jaillet@wanadoo.fr |
|---|---|
| State | New |
| Headers | show |
| Series | mmc: sdhci-sprd: Fix some resource leaks in the remove function | expand |
Le 17/12/2020 à 23:55, Orson Zhai a écrit : > + cc: Billows > > Hi Christophe, > On Fri, Dec 18, 2020 at 4:50 AM Christophe JAILLET > <christophe.jaillet@wanadoo.fr> wrote: >> >> 'sdhci_remove_host()' and 'sdhci_pltfm_free()' should be used in place of >> 'mmc_remove_host()' and 'mmc_free_host()'. >> >> This avoids some resource leaks, is more in line with the error handling >> path of the probe function, and is more consistent with other drivers. >> >> Fixes: fb8bd90f83c4 ("mmc: sdhci-sprd: Add Spreadtrum's initial host controller") >> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> >> --- >> Other adjustment may be needed. >> I'm not sure at all of the 0 passed to 'sdhci_remove_host()'. Some drivers >> pass 0, some have some more complicated computation. >> --- >> drivers/mmc/host/sdhci-sprd.c | 6 +++--- >> 1 file changed, 3 insertions(+), 3 deletions(-) >> >> diff --git a/drivers/mmc/host/sdhci-sprd.c b/drivers/mmc/host/sdhci-sprd.c >> index f85171edabeb..5dc36efff47f 100644 >> --- a/drivers/mmc/host/sdhci-sprd.c >> +++ b/drivers/mmc/host/sdhci-sprd.c >> @@ -708,14 +708,14 @@ static int sdhci_sprd_remove(struct platform_device *pdev) >> { >> struct sdhci_host *host = platform_get_drvdata(pdev); >> struct sdhci_sprd_host *sprd_host = TO_SPRD_HOST(host); >> - struct mmc_host *mmc = host->mmc; >> >> - mmc_remove_host(mmc); >> + sdhci_remove_host(host, 0); >> + >> clk_disable_unprepare(sprd_host->clk_sdio); >> clk_disable_unprepare(sprd_host->clk_enable); >> clk_disable_unprepare(sprd_host->clk_2x_enable); >> >> - mmc_free_host(mmc); >> + sdhci_pltfm_free(pdev); > > I saw a lot of drivers also use mmc_free_host(). > Do you have patches elsewhere to clean them? > As far as I can see, all drivers that use 'mmc_free_host' also use 'mmc_alloc_host'. (based on 5.10.1 and unless error) The only exception is 'sdhci-sprd.c'. So no, I don't plan any other clean-up. To spot it, I run one of my own cocci script which compare functions called in the remove function and in the error handling path of the probe. So I caught this one because 'mmc_free_host' is used in the porbe and 'sdhci_pltfm_free' in the remove function. CJ > Thanks, > -Orson > >> >> return 0; >> } >> -- >> 2.27.0 >> >
On Fri, Dec 18, 2020 at 9:46 PM Christophe JAILLET <christophe.jaillet@wanadoo.fr> wrote: > > Le 17/12/2020 à 23:55, Orson Zhai a écrit : > > + cc: Billows > > > > Hi Christophe, > > On Fri, Dec 18, 2020 at 4:50 AM Christophe JAILLET > > <christophe.jaillet@wanadoo.fr> wrote: > >> > >> 'sdhci_remove_host()' and 'sdhci_pltfm_free()' should be used in place of > >> 'mmc_remove_host()' and 'mmc_free_host()'. > >> > >> This avoids some resource leaks, is more in line with the error handling > >> path of the probe function, and is more consistent with other drivers. > >> > >> Fixes: fb8bd90f83c4 ("mmc: sdhci-sprd: Add Spreadtrum's initial host controller") > >> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> > >> --- > >> Other adjustment may be needed. > >> I'm not sure at all of the 0 passed to 'sdhci_remove_host()'. Some drivers > >> pass 0, some have some more complicated computation. > >> --- > >> drivers/mmc/host/sdhci-sprd.c | 6 +++--- > >> 1 file changed, 3 insertions(+), 3 deletions(-) > >> > >> diff --git a/drivers/mmc/host/sdhci-sprd.c b/drivers/mmc/host/sdhci-sprd.c > >> index f85171edabeb..5dc36efff47f 100644 > >> --- a/drivers/mmc/host/sdhci-sprd.c > >> +++ b/drivers/mmc/host/sdhci-sprd.c > >> @@ -708,14 +708,14 @@ static int sdhci_sprd_remove(struct platform_device *pdev) > >> { > >> struct sdhci_host *host = platform_get_drvdata(pdev); > >> struct sdhci_sprd_host *sprd_host = TO_SPRD_HOST(host); > >> - struct mmc_host *mmc = host->mmc; > >> > >> - mmc_remove_host(mmc); > >> + sdhci_remove_host(host, 0); > >> + > >> clk_disable_unprepare(sprd_host->clk_sdio); > >> clk_disable_unprepare(sprd_host->clk_enable); > >> clk_disable_unprepare(sprd_host->clk_2x_enable); > >> > >> - mmc_free_host(mmc); > >> + sdhci_pltfm_free(pdev); > > > > I saw a lot of drivers also use mmc_free_host(). > > Do you have patches elsewhere to clean them? > > > > As far as I can see, all drivers that use 'mmc_free_host' also use > 'mmc_alloc_host'. (based on 5.10.1 and unless error) > > The only exception is 'sdhci-sprd.c'. > > So no, I don't plan any other clean-up. > > > > To spot it, I run one of my own cocci script which compare functions > called in the remove function and in the error handling path of the probe. > > So I caught this one because 'mmc_free_host' is used in the porbe and > 'sdhci_pltfm_free' in the remove function. Thanks for the clarification. Acked-by: Orson Zhai <orson.zhai@gmail.com> > > > CJ > > > Thanks, > > -Orson > > > >> > >> return 0; > >> } > >> -- > >> 2.27.0 > >> > > >
On 17/12/20 10:42 pm, Christophe JAILLET wrote: > 'sdhci_remove_host()' and 'sdhci_pltfm_free()' should be used in place of > 'mmc_remove_host()' and 'mmc_free_host()'. > > This avoids some resource leaks, is more in line with the error handling > path of the probe function, and is more consistent with other drivers. > > Fixes: fb8bd90f83c4 ("mmc: sdhci-sprd: Add Spreadtrum's initial host controller") > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> Acked-by: Adrian Hunter <adrian.hunter@intel.com> > --- > Other adjustment may be needed. > I'm not sure at all of the 0 passed to 'sdhci_remove_host()'. Some drivers > pass 0, some have some more complicated computation. '0' means cleanup nicely. '1' (dead) means nothing works anymore, including interrupts, so give up on requests immediately. Either should work. '0' is better unless your host controller really is dead, which is mostly not possible except for some legacy PCI devices. > --- > drivers/mmc/host/sdhci-sprd.c | 6 +++--- > 1 file changed, 3 insertions(+), 3 deletions(-) > > diff --git a/drivers/mmc/host/sdhci-sprd.c b/drivers/mmc/host/sdhci-sprd.c > index f85171edabeb..5dc36efff47f 100644 > --- a/drivers/mmc/host/sdhci-sprd.c > +++ b/drivers/mmc/host/sdhci-sprd.c > @@ -708,14 +708,14 @@ static int sdhci_sprd_remove(struct platform_device *pdev) > { > struct sdhci_host *host = platform_get_drvdata(pdev); > struct sdhci_sprd_host *sprd_host = TO_SPRD_HOST(host); > - struct mmc_host *mmc = host->mmc; > > - mmc_remove_host(mmc); > + sdhci_remove_host(host, 0); > + > clk_disable_unprepare(sprd_host->clk_sdio); > clk_disable_unprepare(sprd_host->clk_enable); > clk_disable_unprepare(sprd_host->clk_2x_enable); > > - mmc_free_host(mmc); > + sdhci_pltfm_free(pdev); > > return 0; > } >
On Thu, 17 Dec 2020 at 21:42, Christophe JAILLET <christophe.jaillet@wanadoo.fr> wrote: > > 'sdhci_remove_host()' and 'sdhci_pltfm_free()' should be used in place of > 'mmc_remove_host()' and 'mmc_free_host()'. > > This avoids some resource leaks, is more in line with the error handling > path of the probe function, and is more consistent with other drivers. > > Fixes: fb8bd90f83c4 ("mmc: sdhci-sprd: Add Spreadtrum's initial host controller") > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> Applied for next, thanks! Kind regards Uffe > --- > Other adjustment may be needed. > I'm not sure at all of the 0 passed to 'sdhci_remove_host()'. Some drivers > pass 0, some have some more complicated computation. > --- > drivers/mmc/host/sdhci-sprd.c | 6 +++--- > 1 file changed, 3 insertions(+), 3 deletions(-) > > diff --git a/drivers/mmc/host/sdhci-sprd.c b/drivers/mmc/host/sdhci-sprd.c > index f85171edabeb..5dc36efff47f 100644 > --- a/drivers/mmc/host/sdhci-sprd.c > +++ b/drivers/mmc/host/sdhci-sprd.c > @@ -708,14 +708,14 @@ static int sdhci_sprd_remove(struct platform_device *pdev) > { > struct sdhci_host *host = platform_get_drvdata(pdev); > struct sdhci_sprd_host *sprd_host = TO_SPRD_HOST(host); > - struct mmc_host *mmc = host->mmc; > > - mmc_remove_host(mmc); > + sdhci_remove_host(host, 0); > + > clk_disable_unprepare(sprd_host->clk_sdio); > clk_disable_unprepare(sprd_host->clk_enable); > clk_disable_unprepare(sprd_host->clk_2x_enable); > > - mmc_free_host(mmc); > + sdhci_pltfm_free(pdev); > > return 0; > } > -- > 2.27.0 >
diff --git a/drivers/mmc/host/sdhci-sprd.c b/drivers/mmc/host/sdhci-sprd.c index f85171edabeb..5dc36efff47f 100644 --- a/drivers/mmc/host/sdhci-sprd.c +++ b/drivers/mmc/host/sdhci-sprd.c @@ -708,14 +708,14 @@ static int sdhci_sprd_remove(struct platform_device *pdev) { struct sdhci_host *host = platform_get_drvdata(pdev); struct sdhci_sprd_host *sprd_host = TO_SPRD_HOST(host); - struct mmc_host *mmc = host->mmc; - mmc_remove_host(mmc); + sdhci_remove_host(host, 0); + clk_disable_unprepare(sprd_host->clk_sdio); clk_disable_unprepare(sprd_host->clk_enable); clk_disable_unprepare(sprd_host->clk_2x_enable); - mmc_free_host(mmc); + sdhci_pltfm_free(pdev); return 0; }
'sdhci_remove_host()' and 'sdhci_pltfm_free()' should be used in place of 'mmc_remove_host()' and 'mmc_free_host()'. This avoids some resource leaks, is more in line with the error handling path of the probe function, and is more consistent with other drivers. Fixes: fb8bd90f83c4 ("mmc: sdhci-sprd: Add Spreadtrum's initial host controller") Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> --- Other adjustment may be needed. I'm not sure at all of the 0 passed to 'sdhci_remove_host()'. Some drivers pass 0, some have some more complicated computation. --- drivers/mmc/host/sdhci-sprd.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)