| Message ID | 20201214091616.13545-6-Sergey.Semin@baikalelectronics.ru (mailing list archive) |
|---|---|
| State | Changes Requested |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | net: stmmac: Fix clocks/reset-related procedures | expand |
| Context | Check | Description |
|---|---|---|
| netdev/tree_selection | success | Guessing tree name failed - patch did not apply |
On Mon, 14 Dec 2020 12:15:55 +0300, Serge Semin wrote: > Current clocks description doesn't provide a comprehensive notion about > what "stmmaceth" and "pclk" actually represent from the IP-core manual > point of view. The bindings file states: > stmmaceth - "GMAC main clock", > apb - "Peripheral registers interface clock". > It isn't that easy to understand what they actually mean especially seeing > the DW *MAC manual operates with clock definitions like Application, > System, Host, CSR, Transmit, Receive, etc clocks. Moreover the clocks > usage in the driver doesn't shade a full light on their essence. What > inferred from there is that the "stmmaceth" name has been assigned to the > common clock, which feeds both system and CSR interfaces. But what about > "apb"? The bindings defines it as the clock for "peripheral registers > interface". So it's close to the CSR clock in the IP-core manual notation. > If so then when "apb" clock is specified aside with the "stmmaceth", it > represents a case when the DW *MAC is synthesized with CSR_SLV_CLK=y > (separate system and CSR clocks). But even though the "apb" clock is > requested in the MAC driver, the driver doesn't actually use it as a > separate CSR clock where the IP-core manual requires. All of that makes me > thinking that the case of separate system and CSR clocks isn't correctly > implemented in the driver. > > Let's start with elaborating the clocks description so anyone reading > the DW *MAC bindings file would understand that "stmmaceth" is the > system clock and "pclk" is actually the CSR clock. Indeed in accordance > with sheets depicted in [1]: > system/application clock can be either of: hclk_i, aclk_i, clk_app_i; > CSR clock can be either of: hclk_i, aclk_i, clk_app_i, clk_csr_i. > (Most likely the similar definitions present in the others IP-core > manuals.) So the CSR clock can be tied to the application clock > considering the later as the main clock, but not the other way around. In > case if there is only "stmmaceth" clock specified in a DT node, then it > will be considered as a source of clocks for both application and CSR. But > if "pclk" is also specified in the list of the device clocks, then it will > be perceived as the separate CSR clock. > > [1] DesignWare Cores Ethernet MAC Universal Databook, Revision 3.73a, > October 2013, p. 564. > > Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru> > --- > .../devicetree/bindings/net/snps,dwmac.yaml | 12 ++++++++++-- > 1 file changed, 10 insertions(+), 2 deletions(-) > Reviewed-by: Rob Herring <robh@kernel.org>
diff --git a/Documentation/devicetree/bindings/net/snps,dwmac.yaml b/Documentation/devicetree/bindings/net/snps,dwmac.yaml index 44aa88151cba..e1ebe5c8b1da 100644 --- a/Documentation/devicetree/bindings/net/snps,dwmac.yaml +++ b/Documentation/devicetree/bindings/net/snps,dwmac.yaml @@ -116,8 +116,16 @@ properties: maxItems: 5 additionalItems: true items: - - description: GMAC main clock - - description: Peripheral registers interface clock + - description: + GMAC main clock, also called as system/application clock. + This clock is used to provide a periodic signal for the DMA/MTL + interface and optionally for CSR, if the later isn't separately + clocked. + - description: + Peripheral registers interface clock, also called as CSR clock. + MCI, CSR and SMA interfaces run on this clock. If it's omitted, + the CSR interfaces are considered as synchronous to the system + clock domain. - description: PTP reference clock. This clock is used for programming the Timestamp Addend Register. If not passed then the system
Current clocks description doesn't provide a comprehensive notion about what "stmmaceth" and "pclk" actually represent from the IP-core manual point of view. The bindings file states: stmmaceth - "GMAC main clock", apb - "Peripheral registers interface clock". It isn't that easy to understand what they actually mean especially seeing the DW *MAC manual operates with clock definitions like Application, System, Host, CSR, Transmit, Receive, etc clocks. Moreover the clocks usage in the driver doesn't shade a full light on their essence. What inferred from there is that the "stmmaceth" name has been assigned to the common clock, which feeds both system and CSR interfaces. But what about "apb"? The bindings defines it as the clock for "peripheral registers interface". So it's close to the CSR clock in the IP-core manual notation. If so then when "apb" clock is specified aside with the "stmmaceth", it represents a case when the DW *MAC is synthesized with CSR_SLV_CLK=y (separate system and CSR clocks). But even though the "apb" clock is requested in the MAC driver, the driver doesn't actually use it as a separate CSR clock where the IP-core manual requires. All of that makes me thinking that the case of separate system and CSR clocks isn't correctly implemented in the driver. Let's start with elaborating the clocks description so anyone reading the DW *MAC bindings file would understand that "stmmaceth" is the system clock and "pclk" is actually the CSR clock. Indeed in accordance with sheets depicted in [1]: system/application clock can be either of: hclk_i, aclk_i, clk_app_i; CSR clock can be either of: hclk_i, aclk_i, clk_app_i, clk_csr_i. (Most likely the similar definitions present in the others IP-core manuals.) So the CSR clock can be tied to the application clock considering the later as the main clock, but not the other way around. In case if there is only "stmmaceth" clock specified in a DT node, then it will be considered as a source of clocks for both application and CSR. But if "pclk" is also specified in the list of the device clocks, then it will be perceived as the separate CSR clock. [1] DesignWare Cores Ethernet MAC Universal Databook, Revision 3.73a, October 2013, p. 564. Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru> --- .../devicetree/bindings/net/snps,dwmac.yaml | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-)