Remove git_cpu_init()

[jenswi-linaro]

@Amit-Radur, @b49020, @ChiaweiW, @clementfaure, @ememarar, @etienne-lms, @glneo, @grandpaul, @GseoC, @lorc, @MrVan, @Neal-liu, @pangupta, @sahilnxp, @sdininno

Please check and ack your platforms. One difference for many platforms is that we rely on TF-A or the previous boot stage to initialize the GIC. I assume some unintended copy & paste error has led most platforms to reinitialize the GIC. If I'm wrong in that assumption, we'll see how we can fix it.

[lorc]

For rcar platform:

Reviewed-by: Volodymyr Babchuk <volodymyr_babchuk@epam.com>

[etienne-lms]

Reviewed-by: Etienne Carriere <etienne.carriere@foss.st.com> for commits
"plat-stm: use gic_init_per_cpu()",
"plat-stm32mp1: use gic_init_per_cpu()",
"plat-stm32mp2: use gic_init_per_cpu()" and
"core: remove unused deprecated gic_cpu_init()"
with imperative mode in commit messages :)

[glneo]

Seems fine, but while testing I ran into a panic() on plat-k3, seems to be caused by a different GIC cleanup:

#6394 (comment)

[ChiaweiW]

For aspeed platform:

Reviewed-by: Chia-Wei Wang <chiawei_wang@aspeedtech.com>

[jenswi-linaro]

@glneo, I'm trying to address the issue you mention in #6516

[clementfaure]

I got a crash at boot on old ARMv7 platforms imx-mx6qsabresd

2023-11-30T15:30:43 E/TC:0 0 assertion 'vers == 2' failed at core/drivers/gic.c:469 <gic_init_base_addr>
2023-11-30T15:30:43 E/TC:0 0 Panic at core/kernel/assert.c:28 <_assert_break>
2023-11-30T15:30:43 E/TC:0 0 TEE load address @ 0x14000000
2023-11-30T15:30:43 E/TC:0 0 Call stack:
2023-11-30T15:30:43 E/TC:0 0  0x140039d9
2023-11-30T15:30:43 E/TC:0 0  0x1400c439
2023-11-30T15:30:43 E/TC:0 0  0x14009c03
2023-11-30T15:30:43 E/TC:0 0  0x14008b7b
2023-11-30T15:30:43 E/TC:0 0  0x14009001
2023-11-30T15:30:43 E/TC:0 0  0x14003683
2023-11-30T15:30:43 E/TC:0 0  0x140001fc

I will have a look at it tomorrow

[clementfaure]

Actually, the issue comes from this commit : 69171be where vers = 1 on imx6qsabresd. Cortex-A9 i.MX platforms feature GICv1.

[jenswi-linaro]

@glneo and @clementfaure does the updated #6516 fix your issues?

[clementfaure]

@jenswi-linaro Yes, I can boot the board now (imx8qmmek was also affected).

[sahilnxp]

I was getting the panic while booting on LX2160A-RDB platforms with this pull request.
But after #6516 is merged, board is booting fine.

For commit: plat-ls: use gic_init_per_cpu()
Reviewed-by: Sahil Malhotra <sahil.malhotra@nxp.com>

[jenswi-linaro]

Rebased on master to get the fix from #6516, tags applied and commit messages updated to use imperative mood.

[jenswi-linaro]

A few commits still miss an A-B or R-B tag. I think we can wait a little longer before merging.

[glneo]

For plat-k3: Tested-by: Andrew Davis <afd@ti.com>

For the plat-ti it looks good (not sure if this actually changes anything, so not sure it could break anything), but I haven't gotten around to testing, only a visual code review: Reviewed-by: Andrew Davis <afd@ti.com>

[b49020]

For plat-rzn1: Acked-by: Sumit Garg <sumit.garg@linaro.org>

[jenswi-linaro]

Tags applied. A few commits are still missing tags, but perhaps this is as good as it gets.

[clementfaure]

Tested-by: Clement Faure <clement.faure@nxp.com> (imx-mx6qsabresd, imx-mx6ulevk, imx-mx8qmmek, imx-mx8mnevk)

[jenswi-linaro]

Tag applied.

[jenswi-linaro]

No more progress on tags now. Perhaps this is as good as it gets.

[jforissier]

No more progress on tags now. Perhaps this is as good as it gets.

Yes, I am merging this now. Thanks.