On 3/29/2019 5:15 PM, Marc Zyngier wrote:
On 29/03/2019 11:40, Sameer Pujar wrote:
Hi Marc,
Are you fine with current series?
Please help to get this move forward.
I don't have much objection at the moment, but this is hardly 5.1
material, is it?
I do have audio subsystem patches lined up, which depend on current series.
However I don't know when does next 5.x is going to come and don't think
series is critical for 5.1 at the moment.
Thanks,
Sameer.
Thanks,
M.
Thanks,
Sameer.
On 3/22/2019 5:43 PM, Sameer Pujar wrote:
Currently gic-pm driver is not getting used by any of the Tegra devices.
In fact, the driver is not getting compiled. This series is to ensure,
* build support for driver
* update driver with usage of clk_bulk APIs.
* fix suspend path
This is used by AGIC(Audio Generic Interrupt Controller) block in audio
pipeline.
Changelog
=========
v4->v5:
-------
Following changes are made in current series.
[patch 1/3] arm64: tegra: select ARM_GIC_PM
* no change
[patch 2/3] irqchip/gic-pm: update driver to use clk_bulk APIs
* removed unnecessary null check on 'dev'
* memory for chip_pm is allocated after device match is found
* removed gic_get_clocks() to simplify and probe does required
setup
[patch 3/3] irqchip/gic-pm: fix suspend handling
* no change
v3->v4:
-------
Following changes made in current series.
[patch 1/3] arm64: tegra: select ARM_GIC_PM
* no change
[patch 2/3] irqchip/gic-pm: update driver to use clk_bulk APIs
* commit subject and description changed for clarity
* platform_set_drvdata is replaced by dev_set_drvdata
* NULL check on dev is moved to probe
* replaced devm_kzalloc with devm_kcalloc for chip_pm->clks
* fixed formatting and some cosmetic changes
[patch 3/3] irqchip/gic-pm: fix suspend handling
* no change
v2->v3:
-------
Following are the deltas from previous series
[patch 1/3] arm64: tegra: select ARM_GIC_PM
* no change
[patch 2/3] irqchip/gic-pm: use devm_clk to keep clock state balanced
* commit subject is changed
* 'struct clk_bulk_data' is used instead of 'struct clk' in gic_pm
* clk_bulk() and devm_clk_bulk() APIs are used simplify the code
* Minor changes to line breaks and declaration of variables to make
code more readable
[patch 3/3] irqchip/gic-pm: fix suspend handling
* no change
v1->v2:
-------
following patches are dropped from v1
*[patch 1/5] irqchip/gic-pm: add driver remove support
*[patch 2/5] irqchip/gic: allow gic-pm driver to be used as module
*[patch 3/5] arm64: defconfig: build gic-pm driver as module
The reason for this is, removing gic-pm driver would lead to potential
kernel panic or memory corruption. Thus module option for gic-pm is
rejected.
v2 has 3 patches and the details are below,
[patch 1/3] arm64: tegra: select ARM_GIC_PM
* ARM_GIC_PM is selected for 64-bit Tegra devices
[patch 2/3] irqchip/gic-pm: use devm_clk_*() helpers
* similar to patch 4/5 in v1
[patch 3/3] irqchip/gic-pm: fix suspend handling
* similar to patch 5/5 in v1
v1:
-------
[patch 1/5] irqchip/gic-pm: add driver remove support
* preparatory patch for making gic-pm driver module
* adds remove path in the driver
* gic_teardown() exported from gic driver
[patch 2/5] irqchip/gic: allow gic-pm driver to be used as module
* ARM_GIC_PM config is changed to tristate to allow it to be module
* build errors are fixed by exporting necessary functions from gic
driver
* fixed load and unload errors
[patch 3/5] arm64: defconfig: build gic-pm driver as module
* defconfig change to make driver a kernel module
[patch 4/5] irqchip/gic-pm: use devm_clk_*() helpers
* pm_clk interface is replaced with devm_clk
[patch 5/5] irqchip/gic-pm: fix suspend handling
* system sleep callbacks added to avoid issues during system wide PM
transitions.
Sameer Pujar (3):
arm64: tegra: select ARM_GIC_PM
irqchip/gic-pm: update driver to use clk_bulk APIs
irqchip/gic-pm: fix suspend handling
arch/arm64/Kconfig.platforms | 1 +
drivers/irqchip/irq-gic-pm.c | 76 ++++++++++++++++++++------------------
2 files changed, 40 insertions(+), 37 deletions(-)
