| Message ID | 20230209160704.3791631-3-ioana.ciornei@nxp.com |
|---|---|
| State | Accepted |
| Commit | 8491a7e4ed82a996189a8c97acca90e88f63d345 |
| Delegated to: | Ramon Fried |
| Headers | show |
| Series | drivers: net: fsl-mc: fix MAC address fixup procedure | expand |
On Thu, Feb 9, 2023 at 6:07 PM Ioana Ciornei <ioana.ciornei@nxp.com> wrote: > > In the process of adopting CONFIG_DM_ETH on the DPAA2 based platforms, > interfaces which were previously defined as "xgmii" were transitioned to > be defined as "xfi" in the DTS. > See the commit below for reference: > commit 87274918f2f4 ("arm: dts: ls2088ardb: add DPMAC and PHY nodes") > > Then Vladimir's commit replaced all occurrences of "xfi" with > "10gbase-r" in an effort to make U-Boot work with the same device tree > as Linux. > commit 77b11f760416 ("net: replace the "xfi" phy-mode with "10gbase-r"") > > These changes to the phy_interface_t of an Ethernet port meant that the > mc_fixup_mac_addrs() function was no longer capable to properly fixup > the MAC addresses. The problem arises from the fact that the hardcoded > information about an interface (wriop_get_enet_if()) was no longer > matching any actual device. > > For example, the function tried to search for "DPMAC1@xgmii1" by name > using eth_get_dev_by_name() when only "DPMAC1@10gbase-r" was available. > > This function removes the need to rely on the hardcoded information by > iterating through all the UCLASS_ETH devices which are DPAA2 and request > a fixup for each of them. > > Signed-off-by: Ioana Ciornei <ioana.ciornei@nxp.com> > --- > drivers/net/fsl-mc/mc.c | 31 +++++++++++++------------------ > 1 file changed, 13 insertions(+), 18 deletions(-) > > diff --git a/drivers/net/fsl-mc/mc.c b/drivers/net/fsl-mc/mc.c > index 4f84403d956c..78a40f285aa2 100644 > --- a/drivers/net/fsl-mc/mc.c > +++ b/drivers/net/fsl-mc/mc.c > @@ -29,6 +29,7 @@ > #include <fsl-mc/fsl_dpsparser.h> > #include <fsl-mc/fsl_qbman_portal.h> > #include <fsl-mc/ldpaa_wriop.h> > +#include <net/ldpaa_eth.h> > > #define MC_RAM_BASE_ADDR_ALIGNMENT (512UL * 1024 * 1024) > #define MC_RAM_BASE_ADDR_ALIGNMENT_MASK (~(MC_RAM_BASE_ADDR_ALIGNMENT - 1)) > @@ -383,37 +384,31 @@ static int mc_fixup_dpc_mac_addr(void *blob, int dpmac_id, > > static int mc_fixup_mac_addrs(void *blob, enum mc_fixup_type type) > { > - int i, err = 0, ret = 0; > -#define ETH_NAME_LEN 20 > struct udevice *eth_dev; > - char ethname[ETH_NAME_LEN]; > - > - for (i = WRIOP1_DPMAC1; i < NUM_WRIOP_PORTS; i++) { > - /* port not enabled */ > - if (wriop_is_enabled_dpmac(i) != 1) > - continue; > - > - snprintf(ethname, ETH_NAME_LEN, "DPMAC%d@%s", i, > - phy_interface_strings[wriop_get_enet_if(i)]); > - > - eth_dev = eth_get_dev_by_name(ethname); > - if (eth_dev == NULL) > + int err = 0, ret = 0; > + struct uclass *uc; > + uint32_t dpmac_id; > + > + uclass_get(UCLASS_ETH, &uc); > + uclass_foreach_dev(eth_dev, uc) { > + if (!eth_dev->driver || !eth_dev->driver->name || > + strcmp(eth_dev->driver->name, LDPAA_ETH_DRIVER_NAME)) > continue; > > + dpmac_id = ldpaa_eth_get_dpmac_id(eth_dev); > switch (type) { > case MC_FIXUP_DPL: > - err = mc_fixup_dpl_mac_addr(blob, i, eth_dev); > + err = mc_fixup_dpl_mac_addr(blob, dpmac_id, eth_dev); > break; > case MC_FIXUP_DPC: > - err = mc_fixup_dpc_mac_addr(blob, i, eth_dev); > + err = mc_fixup_dpc_mac_addr(blob, dpmac_id, eth_dev); > break; > default: > break; > } > > if (err) > - printf("fsl-mc: ERROR fixing mac address for %s\n", > - ethname); > + printf("fsl-mc: ERROR fixing mac address for %s\n", eth_dev->name); > ret |= err; > } > > -- > 2.25.1 > Reviewed-by: Ramon Fried <rfried.dev@gmail.com>
On Thu, Feb 09, 2023 at 06:07:04PM +0200, Ioana Ciornei wrote: > In the process of adopting CONFIG_DM_ETH on the DPAA2 based platforms, > interfaces which were previously defined as "xgmii" were transitioned to > be defined as "xfi" in the DTS. > See the commit below for reference: > commit 87274918f2f4 ("arm: dts: ls2088ardb: add DPMAC and PHY nodes") > > Then Vladimir's commit replaced all occurrences of "xfi" with > "10gbase-r" in an effort to make U-Boot work with the same device tree > as Linux. > commit 77b11f760416 ("net: replace the "xfi" phy-mode with "10gbase-r"") > > These changes to the phy_interface_t of an Ethernet port meant that the > mc_fixup_mac_addrs() function was no longer capable to properly fixup > the MAC addresses. The problem arises from the fact that the hardcoded > information about an interface (wriop_get_enet_if()) was no longer > matching any actual device. > > For example, the function tried to search for "DPMAC1@xgmii1" by name > using eth_get_dev_by_name() when only "DPMAC1@10gbase-r" was available. > > This function removes the need to rely on the hardcoded information by > iterating through all the UCLASS_ETH devices which are DPAA2 and request > a fixup for each of them. > > Signed-off-by: Ioana Ciornei <ioana.ciornei@nxp.com> > Reviewed-by: Ramon Fried <rfried.dev@gmail.com> Applied to u-boot/master, thanks!
diff --git a/drivers/net/fsl-mc/mc.c b/drivers/net/fsl-mc/mc.c index 4f84403d956c..78a40f285aa2 100644 --- a/drivers/net/fsl-mc/mc.c +++ b/drivers/net/fsl-mc/mc.c @@ -29,6 +29,7 @@ #include <fsl-mc/fsl_dpsparser.h> #include <fsl-mc/fsl_qbman_portal.h> #include <fsl-mc/ldpaa_wriop.h> +#include <net/ldpaa_eth.h> #define MC_RAM_BASE_ADDR_ALIGNMENT (512UL * 1024 * 1024) #define MC_RAM_BASE_ADDR_ALIGNMENT_MASK (~(MC_RAM_BASE_ADDR_ALIGNMENT - 1)) @@ -383,37 +384,31 @@ static int mc_fixup_dpc_mac_addr(void *blob, int dpmac_id, static int mc_fixup_mac_addrs(void *blob, enum mc_fixup_type type) { - int i, err = 0, ret = 0; -#define ETH_NAME_LEN 20 struct udevice *eth_dev; - char ethname[ETH_NAME_LEN]; - - for (i = WRIOP1_DPMAC1; i < NUM_WRIOP_PORTS; i++) { - /* port not enabled */ - if (wriop_is_enabled_dpmac(i) != 1) - continue; - - snprintf(ethname, ETH_NAME_LEN, "DPMAC%d@%s", i, - phy_interface_strings[wriop_get_enet_if(i)]); - - eth_dev = eth_get_dev_by_name(ethname); - if (eth_dev == NULL) + int err = 0, ret = 0; + struct uclass *uc; + uint32_t dpmac_id; + + uclass_get(UCLASS_ETH, &uc); + uclass_foreach_dev(eth_dev, uc) { + if (!eth_dev->driver || !eth_dev->driver->name || + strcmp(eth_dev->driver->name, LDPAA_ETH_DRIVER_NAME)) continue; + dpmac_id = ldpaa_eth_get_dpmac_id(eth_dev); switch (type) { case MC_FIXUP_DPL: - err = mc_fixup_dpl_mac_addr(blob, i, eth_dev); + err = mc_fixup_dpl_mac_addr(blob, dpmac_id, eth_dev); break; case MC_FIXUP_DPC: - err = mc_fixup_dpc_mac_addr(blob, i, eth_dev); + err = mc_fixup_dpc_mac_addr(blob, dpmac_id, eth_dev); break; default: break; } if (err) - printf("fsl-mc: ERROR fixing mac address for %s\n", - ethname); + printf("fsl-mc: ERROR fixing mac address for %s\n", eth_dev->name); ret |= err; }
In the process of adopting CONFIG_DM_ETH on the DPAA2 based platforms, interfaces which were previously defined as "xgmii" were transitioned to be defined as "xfi" in the DTS. See the commit below for reference: commit 87274918f2f4 ("arm: dts: ls2088ardb: add DPMAC and PHY nodes") Then Vladimir's commit replaced all occurrences of "xfi" with "10gbase-r" in an effort to make U-Boot work with the same device tree as Linux. commit 77b11f760416 ("net: replace the "xfi" phy-mode with "10gbase-r"") These changes to the phy_interface_t of an Ethernet port meant that the mc_fixup_mac_addrs() function was no longer capable to properly fixup the MAC addresses. The problem arises from the fact that the hardcoded information about an interface (wriop_get_enet_if()) was no longer matching any actual device. For example, the function tried to search for "DPMAC1@xgmii1" by name using eth_get_dev_by_name() when only "DPMAC1@10gbase-r" was available. This function removes the need to rely on the hardcoded information by iterating through all the UCLASS_ETH devices which are DPAA2 and request a fixup for each of them. Signed-off-by: Ioana Ciornei <ioana.ciornei@nxp.com> --- drivers/net/fsl-mc/mc.c | 31 +++++++++++++------------------ 1 file changed, 13 insertions(+), 18 deletions(-)