| Message ID | 20241218012849.3214468-1-joe@pf.is.s.u-tokyo.ac.jp (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [v2] net: mv643xx_eth: fix an OF node reference leak | expand |
On 12/18/24 02:28, Joe Hattori wrote: > Current implementation of mv643xx_eth_shared_of_add_port() calls > of_parse_phandle(), but does not release the refcount on error. Call > of_node_put() in the error path and in mv643xx_eth_shared_of_remove(). > > This bug was found by an experimental verification tool that I am > developing. > > Fixes: 76723bca2802 ("net: mv643xx_eth: add DT parsing support") > Signed-off-by: Joe Hattori <joe@pf.is.s.u-tokyo.ac.jp> > --- > Changes in v3: > - Insert a NULL check for port_platdev[n]. > Changes in v2: > - Insert a NULL check before accessing the platform data. I'm sorry for nit-picking, but many little things are adding-up and should be noticed. The subj prefix must include the correct revision number (v3 in this case). You must avoid submitting new revisions within the 24h grace period, see: https://elixir.bootlin.com/linux/v6.12.5/source/Documentation/process/maintainer-netdev.rst#L414 > --- > drivers/net/ethernet/marvell/mv643xx_eth.c | 14 ++++++++++++-- > 1 file changed, 12 insertions(+), 2 deletions(-) > > diff --git a/drivers/net/ethernet/marvell/mv643xx_eth.c b/drivers/net/ethernet/marvell/mv643xx_eth.c > index a06048719e84..0e2ebfcaad1c 100644 > --- a/drivers/net/ethernet/marvell/mv643xx_eth.c > +++ b/drivers/net/ethernet/marvell/mv643xx_eth.c > @@ -2705,8 +2705,14 @@ static struct platform_device *port_platdev[3]; > static void mv643xx_eth_shared_of_remove(void) > { > int n; > + struct mv643xx_eth_platform_data *pd; Please respect the reverse xmas tree above, see: https://elixir.bootlin.com/linux/v6.12.5/source/Documentation/process/maintainer-netdev.rst#L360 Otherwise LGTM, thanks! Paolo
On 12/19/24 11:21, Paolo Abeni wrote: > On 12/18/24 02:28, Joe Hattori wrote: >> Current implementation of mv643xx_eth_shared_of_add_port() calls >> of_parse_phandle(), but does not release the refcount on error. Call >> of_node_put() in the error path and in mv643xx_eth_shared_of_remove(). >> >> This bug was found by an experimental verification tool that I am >> developing. >> >> Fixes: 76723bca2802 ("net: mv643xx_eth: add DT parsing support") >> Signed-off-by: Joe Hattori <joe@pf.is.s.u-tokyo.ac.jp> >> --- >> Changes in v3: >> - Insert a NULL check for port_platdev[n]. >> Changes in v2: >> - Insert a NULL check before accessing the platform data. > > I'm sorry for nit-picking, but many little things are adding-up and > should be noticed. > > The subj prefix must include the correct revision number (v3 in this case). Oops, I almost forgot... the patch subj prefix must additionally include the target tree - 'net' in this case - see: https://elixir.bootlin.com/linux/v6.12.5/source/Documentation/process/maintainer-netdev.rst#L332 Thanks, Paolo
Thank you for your review. On 12/19/24 19:23, Paolo Abeni wrote: > On 12/19/24 11:21, Paolo Abeni wrote: >> On 12/18/24 02:28, Joe Hattori wrote: >>> Current implementation of mv643xx_eth_shared_of_add_port() calls >>> of_parse_phandle(), but does not release the refcount on error. Call >>> of_node_put() in the error path and in mv643xx_eth_shared_of_remove(). >>> >>> This bug was found by an experimental verification tool that I am >>> developing. >>> >>> Fixes: 76723bca2802 ("net: mv643xx_eth: add DT parsing support") >>> Signed-off-by: Joe Hattori <joe@pf.is.s.u-tokyo.ac.jp> >>> --- >>> Changes in v3: >>> - Insert a NULL check for port_platdev[n]. >>> Changes in v2: >>> - Insert a NULL check before accessing the platform data. >> >> I'm sorry for nit-picking, but many little things are adding-up and >> should be noticed. >> >> The subj prefix must include the correct revision number (v3 in this case). > > Oops, I almost forgot... the patch subj prefix must additionally include > the target tree - 'net' in this case - see: > > https://elixir.bootlin.com/linux/v6.12.5/source/Documentation/process/maintainer-netdev.rst#L332 Thank you for pointing out. All the points you raised have been fixed in the v4 patch. > > Thanks, > > Paolo > Best, Joe
diff --git a/drivers/net/ethernet/marvell/mv643xx_eth.c b/drivers/net/ethernet/marvell/mv643xx_eth.c index a06048719e84..0e2ebfcaad1c 100644 --- a/drivers/net/ethernet/marvell/mv643xx_eth.c +++ b/drivers/net/ethernet/marvell/mv643xx_eth.c @@ -2705,8 +2705,14 @@ static struct platform_device *port_platdev[3]; static void mv643xx_eth_shared_of_remove(void) { int n; + struct mv643xx_eth_platform_data *pd; for (n = 0; n < 3; n++) { + if (!port_platdev[n]) + continue; + pd = dev_get_platdata(&port_platdev[n]->dev); + if (pd) + of_node_put(pd->phy_node); platform_device_del(port_platdev[n]); port_platdev[n] = NULL; } @@ -2769,8 +2775,10 @@ static int mv643xx_eth_shared_of_add_port(struct platform_device *pdev, } ppdev = platform_device_alloc(MV643XX_ETH_NAME, dev_num); - if (!ppdev) - return -ENOMEM; + if (!ppdev) { + ret = -ENOMEM; + goto put_err; + } ppdev->dev.coherent_dma_mask = DMA_BIT_MASK(32); ppdev->dev.of_node = pnp; @@ -2792,6 +2800,8 @@ static int mv643xx_eth_shared_of_add_port(struct platform_device *pdev, port_err: platform_device_put(ppdev); +put_err: + of_node_put(ppd.phy_node); return ret; }
Current implementation of mv643xx_eth_shared_of_add_port() calls of_parse_phandle(), but does not release the refcount on error. Call of_node_put() in the error path and in mv643xx_eth_shared_of_remove(). This bug was found by an experimental verification tool that I am developing. Fixes: 76723bca2802 ("net: mv643xx_eth: add DT parsing support") Signed-off-by: Joe Hattori <joe@pf.is.s.u-tokyo.ac.jp> --- Changes in v3: - Insert a NULL check for port_platdev[n]. Changes in v2: - Insert a NULL check before accessing the platform data. --- drivers/net/ethernet/marvell/mv643xx_eth.c | 14 ++++++++++++-- 1 file changed, 12 insertions(+), 2 deletions(-)