| Message ID | 20240816073908.2343650-1-b-padhi@ti.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | Reserve Timer Nodes to avoid clash with Remoteproc firmware | expand |
On 13:09-20240816, Beleswar Padhi wrote: > The remoteproc firmware like of R5F and DSPs in the MAIN voltage domain > use timers. At the same time, if Linux probes the timers, some > instability is observed while booting remote cores. Therefore, change > the status of the timer nodes to "reserved" to avoid any clash. > > v3: Changelog: > * Udit > 1) j7200: Updated commit message description to remove DSPs as the SoC does not > have one. > 2) j722s: Updated the correct commit for "Fixes:" tag which introduced the bug. > 3) For entire series, removed comment about timer usage by R5fs in the MCU > voltage domain as those are used by Device Manager firmware to get SoC > functional. > > Link to v2: > https://lore.kernel.org/all/20240814104151.2038457-1-b-padhi@ti.com/ > > v2: Changelog: > - Split the changes into individual patches for each SoC to tag the > correct offending commit for "Fixes:" tag. > > * Udit > 1) Add the correct clashing timer nodes for J7200 SoC. > 2) Port these changes to board level dts files instead of SoC level dtsi files. > > Link to v1: > https://lore.kernel.org/all/20240607105559.771080-1-b-padhi@ti.com/ > > Beleswar Padhi (5): > arm64: dts: ti: k3-j7200: Change timer nodes status to reserved > arm64: dts: ti: k3-j721e: Change timer nodes status to reserved > arm64: dts: ti: k3-j721s2: Change timer nodes status to reserved > arm64: dts: ti: k3-j784s4: Change timer nodes status to reserved > arm64: dts: ti: k3-j722s: Change timer nodes status to reserved All patches: please fix your $subject. it looked initially like you were fixing the default SoC behavior, but later I realized you are fixing the evm solution. Also while at this make sure the corresponding SK behaviors are in sync correctly as well. if you think they do not need to be fixed, address that comment in the cover-letter. > > .../dts/ti/k3-j7200-common-proc-board.dts | 13 ++++++ > .../dts/ti/k3-j721e-common-proc-board.dts | 29 +++++++++++++ > .../dts/ti/k3-j721s2-common-proc-board.dts | 25 +++++++++++ > arch/arm64/boot/dts/ti/k3-j722s-evm.dts | 13 ++++++ > arch/arm64/boot/dts/ti/k3-j784s4-evm.dts | 41 +++++++++++++++++++ > 5 files changed, 121 insertions(+) > > -- > 2.34.1 >
On 24/08/24 23:33, Nishanth Menon wrote: > On 13:09-20240816, Beleswar Padhi wrote: >> The remoteproc firmware like of R5F and DSPs in the MAIN voltage domain >> use timers. At the same time, if Linux probes the timers, some >> instability is observed while booting remote cores. Therefore, change >> the status of the timer nodes to "reserved" to avoid any clash. >> >> v3: Changelog: >> * Udit >> 1) j7200: Updated commit message description to remove DSPs as the SoC does not >> have one. >> 2) j722s: Updated the correct commit for "Fixes:" tag which introduced the bug. >> 3) For entire series, removed comment about timer usage by R5fs in the MCU >> voltage domain as those are used by Device Manager firmware to get SoC >> functional. >> >> Link to v2: >> https://lore.kernel.org/all/20240814104151.2038457-1-b-padhi@ti.com/ >> >> v2: Changelog: >> - Split the changes into individual patches for each SoC to tag the >> correct offending commit for "Fixes:" tag. >> >> * Udit >> 1) Add the correct clashing timer nodes for J7200 SoC. >> 2) Port these changes to board level dts files instead of SoC level dtsi files. >> >> Link to v1: >> https://lore.kernel.org/all/20240607105559.771080-1-b-padhi@ti.com/ >> >> Beleswar Padhi (5): >> arm64: dts: ti: k3-j7200: Change timer nodes status to reserved >> arm64: dts: ti: k3-j721e: Change timer nodes status to reserved >> arm64: dts: ti: k3-j721s2: Change timer nodes status to reserved >> arm64: dts: ti: k3-j784s4: Change timer nodes status to reserved >> arm64: dts: ti: k3-j722s: Change timer nodes status to reserved > All patches: please fix your $subject. it looked initially like Thanks for pointing out, fixed this in v4 revision. > you were fixing the default SoC behavior, but later I realized > you are fixing the evm solution. > > Also while at this make sure the corresponding SK behaviors are in sync Thanks for catching this. I have included timer changes for SK boards in the v4 revision. > correctly as well. if you think they do not need to be fixed, address > that comment in the cover-letter. > >> .../dts/ti/k3-j7200-common-proc-board.dts | 13 ++++++ >> .../dts/ti/k3-j721e-common-proc-board.dts | 29 +++++++++++++ >> .../dts/ti/k3-j721s2-common-proc-board.dts | 25 +++++++++++ >> arch/arm64/boot/dts/ti/k3-j722s-evm.dts | 13 ++++++ >> arch/arm64/boot/dts/ti/k3-j784s4-evm.dts | 41 +++++++++++++++++++ >> 5 files changed, 121 insertions(+) >> >> -- >> 2.34.1 >>