| Message ID | 20200701165128.1213447-1-codrin.ciubotariu@microchip.com |
|---|---|
| State | Changes Requested |
| Delegated to: | David Miller |
| Headers | show |
| Series | [1/2] net: dsa: microchip: set the correct number of ports in dsa_switch | expand |
On Wed, Jul 01, 2020 at 07:51:27PM +0300, Codrin Ciubotariu wrote: > The number of ports is incorrectly set to the maximum available for a DSA > switch. Even if the extra ports are not used, this causes some functions > to be called later, like port_disable() and port_stp_state_set(). If the > driver doesn't check the port index, it will end up modifying unknown > registers. > > Fixes: b987e98e50ab ("dsa: add DSA switch driver for Microchip KSZ9477") Hi Codrin You don't indicate which tree this is for. net-next, or net? It looks like it fixes a real issue, so it probably should be for net. But patches to net should be minimal. Is it possible to do the ds->num_ports = swdev->port_cnt; without all the other changes? You can then have a refactoring patch in net-next. Thanks Andrew
On 01.07.2020 20:09, Andrew Lunn wrote: > EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe > > On Wed, Jul 01, 2020 at 07:51:27PM +0300, Codrin Ciubotariu wrote: >> The number of ports is incorrectly set to the maximum available for a DSA >> switch. Even if the extra ports are not used, this causes some functions >> to be called later, like port_disable() and port_stp_state_set(). If the >> driver doesn't check the port index, it will end up modifying unknown >> registers. >> >> Fixes: b987e98e50ab ("dsa: add DSA switch driver for Microchip KSZ9477") > > Hi Codrin > > You don't indicate which tree this is for. net-next, or net? It looks > like it fixes a real issue, so it probably should be for net. But > patches to net should be minimal. Is it possible to do the > > ds->num_ports = swdev->port_cnt; > > without all the other changes? You can then have a refactoring patch > in net-next. This one should be for net. Ok then, I will send a simpler version of this patch for net, just to fix the issue and another one like this one for net-next. Thanks and best regards, Codrin > > Thanks > Andrew >
diff --git a/drivers/net/dsa/microchip/ksz8795.c b/drivers/net/dsa/microchip/ksz8795.c index 47d65b77caf7..b0227b0e31e6 100644 --- a/drivers/net/dsa/microchip/ksz8795.c +++ b/drivers/net/dsa/microchip/ksz8795.c @@ -1225,8 +1225,6 @@ static int ksz8795_switch_init(struct ksz_device *dev) { int i; - dev->ds->ops = &ksz8795_switch_ops; - for (i = 0; i < ARRAY_SIZE(ksz8795_switch_chips); i++) { const struct ksz_chip_data *chip = &ksz8795_switch_chips[i]; @@ -1268,6 +1266,11 @@ static int ksz8795_switch_init(struct ksz_device *dev) return -ENOMEM; } + dev->ds = ksz_dsa_switch_alloc(dev); + if (!dev->ds) + return -ENOMEM; + dev->ds->ops = &ksz8795_switch_ops; + return 0; } diff --git a/drivers/net/dsa/microchip/ksz9477.c b/drivers/net/dsa/microchip/ksz9477.c index 9a51b8a4de5d..833cf3763000 100644 --- a/drivers/net/dsa/microchip/ksz9477.c +++ b/drivers/net/dsa/microchip/ksz9477.c @@ -1545,8 +1545,6 @@ static int ksz9477_switch_init(struct ksz_device *dev) { int i; - dev->ds->ops = &ksz9477_switch_ops; - for (i = 0; i < ARRAY_SIZE(ksz9477_switch_chips); i++) { const struct ksz_chip_data *chip = &ksz9477_switch_chips[i]; @@ -1588,6 +1586,11 @@ static int ksz9477_switch_init(struct ksz_device *dev) return -ENOMEM; } + dev->ds = ksz_dsa_switch_alloc(dev); + if (!dev->ds) + return -ENOMEM; + dev->ds->ops = &ksz9477_switch_ops; + return 0; } diff --git a/drivers/net/dsa/microchip/ksz_common.c b/drivers/net/dsa/microchip/ksz_common.c index fd1d6676ae4f..4a41f0c7dbbc 100644 --- a/drivers/net/dsa/microchip/ksz_common.c +++ b/drivers/net/dsa/microchip/ksz_common.c @@ -387,30 +387,36 @@ EXPORT_SYMBOL_GPL(ksz_disable_port); struct ksz_device *ksz_switch_alloc(struct device *base, void *priv) { - struct dsa_switch *ds; struct ksz_device *swdev; - ds = devm_kzalloc(base, sizeof(*ds), GFP_KERNEL); - if (!ds) - return NULL; - - ds->dev = base; - ds->num_ports = DSA_MAX_PORTS; - swdev = devm_kzalloc(base, sizeof(*swdev), GFP_KERNEL); if (!swdev) return NULL; - ds->priv = swdev; swdev->dev = base; - swdev->ds = ds; swdev->priv = priv; return swdev; } EXPORT_SYMBOL(ksz_switch_alloc); +struct dsa_switch *ksz_dsa_switch_alloc(struct ksz_device *swdev) +{ + struct dsa_switch *ds; + + ds = devm_kzalloc(swdev->dev, sizeof(*ds), GFP_KERNEL); + if (!ds) + return NULL; + + ds->dev = swdev->dev; + ds->num_ports = swdev->port_cnt; + ds->priv = swdev; + + return ds; +} +EXPORT_SYMBOL(ksz_dsa_switch_alloc); + int ksz_switch_register(struct ksz_device *dev, const struct ksz_dev_ops *ops) { diff --git a/drivers/net/dsa/microchip/ksz_common.h b/drivers/net/dsa/microchip/ksz_common.h index f2c9bb68fd33..785702514a46 100644 --- a/drivers/net/dsa/microchip/ksz_common.h +++ b/drivers/net/dsa/microchip/ksz_common.h @@ -145,6 +145,7 @@ struct ksz_dev_ops { }; struct ksz_device *ksz_switch_alloc(struct device *base, void *priv); +struct dsa_switch *ksz_dsa_switch_alloc(struct ksz_device *swdev); int ksz_switch_register(struct ksz_device *dev, const struct ksz_dev_ops *ops); void ksz_switch_remove(struct ksz_device *dev);
The number of ports is incorrectly set to the maximum available for a DSA switch. Even if the extra ports are not used, this causes some functions to be called later, like port_disable() and port_stp_state_set(). If the driver doesn't check the port index, it will end up modifying unknown registers. Fixes: b987e98e50ab ("dsa: add DSA switch driver for Microchip KSZ9477") Signed-off-by: Codrin Ciubotariu <codrin.ciubotariu@microchip.com> --- drivers/net/dsa/microchip/ksz8795.c | 7 +++++-- drivers/net/dsa/microchip/ksz9477.c | 7 +++++-- drivers/net/dsa/microchip/ksz_common.c | 26 ++++++++++++++++---------- drivers/net/dsa/microchip/ksz_common.h | 1 + 4 files changed, 27 insertions(+), 14 deletions(-)