Add the new platform ACP_7_0 #9351

saisurya-ch · 2024-08-07T09:32:50Z

Add the new platform support for ACP_7_0.
Add built support for xt-clang compiler.

As already discussed in commit bcbcec7 ("cmake: drop binutils-specific '-Wl,-EL' option") and commit ee58fef ("smex/cmake: move -Wl,EL option to target_linker_options() for clang") and their corresponding reviews on GitHub, this binutils-specific, endianness option makes no difference and is causing clang compatibility issues. It's now getting in the way of thesofproject#9351 "Add the new platform ACP_7_0". I ran `./scripts/docker-run.sh ./scripts/xtensa-build-all.sh -a` with and without it and there was absolutely zero binary difference. I was added in 2019 by giant commit e0ba98d ("cmake: add CMakeLists for firmware build") without any rationale of why it would be needed. Signed-off-by: Marc Herbert <[email protected]>

marc-hb

Some changes requested in build scripts, see below.

As usual I was curious about the copy/paste/diverge ratio so I quickly copied src/platform/amd/acp_7_0/ over src/platform/amd/acp_6_3/ and ran git diff --stat. The duplication seems to be at least 85%

$ git diff --stat
 src/platform/amd/acp_6_3/include/arch/xtensa/config/core-isa.h    | 160 +++++++++++++++++++++++++++++++-----------------
 src/platform/amd/acp_6_3/include/arch/xtensa/config/core-matmap.h | 282 ++++++++++---------------------------------------------------------------------------
 src/platform/amd/acp_6_3/include/arch/xtensa/config/defs.h        |   2 +-
 src/platform/amd/acp_6_3/include/arch/xtensa/config/specreg.h     |  11 ++--
 src/platform/amd/acp_6_3/include/arch/xtensa/config/system.h      |   6 +-
 src/platform/amd/acp_6_3/include/arch/xtensa/config/tie-asm.h     |  42 +++++++++----
 src/platform/amd/acp_6_3/include/arch/xtensa/config/tie.h         |  52 ++++++++--------
 src/platform/amd/acp_6_3/include/arch/xtensa/tie/xt_datacache.h   |  61 +++++++++----------
 src/platform/amd/acp_6_3/include/platform/chip_offset_byte.h      | 106 +++++++++++++++-----------------
 src/platform/amd/acp_6_3/include/platform/chip_registers.h        |  85 +++++++++++++-------------
 src/platform/amd/acp_6_3/include/platform/drivers/interrupt.h     |   5 +-
 src/platform/amd/acp_6_3/include/platform/fw_scratch_mem.h        |   7 +--
 src/platform/amd/acp_6_3/include/platform/lib/memory.h            |  11 ++--
 src/platform/amd/acp_6_3/include/platform/platform.h              |   5 +-
 src/platform/amd/acp_6_3/lib/clk.c                                | 235 +++++++++++++++++++++++++++++++++++------------------------------------
 src/platform/amd/acp_6_3/lib/dai.c                                | 104 +------------------------------
 src/platform/amd/acp_6_3/lib/dma.c                                |  21 +------
 src/platform/amd/acp_6_3/lib/memory.c                             |   5 +-
 src/platform/amd/acp_6_3/platform.c                               |   5 +-
 19 files changed, 459 insertions(+), 746 deletions(-)

wc -l $(find acp_6_3/ -type f | sort)

   581 acp_6_3/acp_6_3.x.in
   586 acp_6_3/acp_7_0.x.in
     5 acp_6_3/CMakeLists.txt
   714 acp_6_3/include/arch/xtensa/config/core-isa.h
   317 acp_6_3/include/arch/xtensa/config/core-matmap.h
    38 acp_6_3/include/arch/xtensa/config/defs.h
   108 acp_6_3/include/arch/xtensa/config/specreg.h
   262 acp_6_3/include/arch/xtensa/config/system.h
   366 acp_6_3/include/arch/xtensa/config/tie-asm.h
   211 acp_6_3/include/arch/xtensa/config/tie.h
   132 acp_6_3/include/arch/xtensa/tie/xt_datacache.h
   230 acp_6_3/include/platform/chip_offset_byte.h
  1062 acp_6_3/include/platform/chip_registers.h
   109 acp_6_3/include/platform/drivers/interrupt.h
   120 acp_6_3/include/platform/fw_scratch_mem.h
   185 acp_6_3/include/platform/lib/memory.h
   101 acp_6_3/include/platform/platform.h
   493 acp_6_3/lib/clk.c
     8 acp_6_3/lib/CMakeLists.txt
   296 acp_6_3/lib/dai.c
   150 acp_6_3/lib/dma.c
    95 acp_6_3/lib/memory.c
   210 acp_6_3/platform.c

  6379 total

marc-hb · 2024-08-07T15:47:53Z

scripts/xtensa-build-all.sh

+SUPPORTED_PLATFORMS+=( acp_6_3 acp_7_0 )
+
+# Platforms support xt-clang compiler build
+CLANG_BUILD_PLATFORMS=( acp_7_0 )


Instead of adding this new variable, define XCC at the bottom of scripts/set_xtensa_params.sh. Copy and modify the ZEPHYR_TOOLCHAIN_VARIANT code which is already there.

# for xtensa-build-all.sh case "$platform" in acp_7_0) XCC='xt-clang';; *) XCC='xcc';; esac

Have an issue If setting XCC this way: In the build case of acp_7_0 GCC build (where XTENSA_TOOLS_ROOT is empty), XCC sets to 'xt-clang' and later in file "scripts/xtensa-build-all.sh" at line 228 if [ "$XCC" == "xt-xcc" ] || [ "$XCC" == "xt-clang" ] gets true and sets COMPILER="clang" which is wrong. Actually need to set the COMPILER="gcc" for which the XCC need to be empty.

Thanks, this required a bit more clean-up than I expected. I did it in #9358. That should work for you, please test and review.

marc-hb · 2024-08-07T15:56:18Z

scripts/xtensa-build-all.sh

@@ -253,6 +266,7 @@ do
 	printf 'PATH=%s\n' "$PATH"
 	( set -x # log the main commands and their parameters
 	cmake -DTOOLCHAIN="$TOOLCHAIN" \
+		-DCOMPILER="$COMPILER" \


The COMPILER name is not specific enough, see above.

made it -DSOF_XT_COMPILER="$COMPILER" \

marc-hb · 2024-08-07T15:58:18Z

src/arch/xtensa/CMakeLists.txt

+		>
+		-imacros${CONFIG_H_PATH}
+		)
+else()


Don't copy/paste/diverge 10 lines to make a single line difference (which took me way too long to spot).

If this difference was needed (it's not, see below) then you would do something like:

-Wall -Werror ${EL_option} -Wmissing-prototypes

BUT, even simpler: I'm pretty sure this option is not needed AT ALL. Please try simply removing it and if that works ACP_7_0 then approve small #9353 which I just submitted and which should be merged first. Then rebase on top of it.

yes, removing -Wl,-EL works fine.

Please wait until the code is actually changed and pushed before "resolving" conversations, otherwise it looks like this particular change is OK (it's not).

(Some people don't "resolve" conversations even after the code has changed. It's not required by this project and it does not really matter except when there are too many of them.)

marc-hb · 2024-08-07T17:03:06Z

scripts/cmake/xtensa-toolchain.cmake

@@ -49,7 +49,7 @@ set(CMAKE_TRY_COMPILE_TARGET_TYPE STATIC_LIBRARY)
 # xt toolchain only partially follows gcc convention
 if(TOOLCHAIN STREQUAL "xt")
 	set(XCC 1)
-	set(CMAKE_C_COMPILER ${CROSS_COMPILE}xcc)
+	set(CMAKE_C_COMPILER ${CROSS_COMPILE}${COMPILER})


COMPILER is an extremely common name, very high risk of conflicts and impossible to git grep. Use a more specific name, maybe: SOF_XT_COMPILER

marc-hb · 2024-08-07T17:04:11Z

scripts/xtensa-build-all.sh

+					XCC="xt-clang"
+				else
+					XCC="xt-xcc"
+				fi


marc-hb · 2024-08-07T17:06:29Z

src/arch/xtensa/include/arch/spinlock.h

 	lock->lock = 0;
-	result = 1;


The code change looks good but the commit message does not:

Remove the unused variable 'result' that is causing the GCC build error.

Which error? Which gcc? This has been successfully compiled by gcc since forever.

[ 5%] Building ASM object src/arch/xtensa/xtos/CMakeFiles/xlevel4.dir/int-vector.S.o [ 7%] Building ASM object src/arch/xtensa/xtos/CMakeFiles/xlevel4.dir/int-initlevel.S.o [ 7%] Building ASM object src/arch/xtensa/xtos/CMakeFiles/xlevel3.dir/int-handler.S.o [ 8%] Building ASM object src/arch/xtensa/hal/CMakeFiles/hal.dir/int_asm.S.o [ 8%] Building ASM object src/arch/xtensa/hal/CMakeFiles/hal.dir/clock.S.o [ 8%] Building C object src/arch/xtensa/hal/CMakeFiles/hal.dir/interrupts.c.o [ 10%] Building ASM object src/arch/xtensa/xtos/CMakeFiles/xlevel5.dir/int-handler.S.o [ 10%] Building ASM object src/arch/xtensa/xtos/CMakeFiles/xlevel3.dir/int-vector.S.o [ 11%] No download step for 'smex_ep' In file included from /home/sof/xtos/include/rtos/spinlock.h:16, from /home/sof/xtos/include/rtos/sof.h:14, from /home/sof/src/include/sof/debug/debug.h:16, from /home/sof/src/init/ext_manifest.c:11: /home/sof/src/arch/xtensa/include/arch/spinlock.h: In function 'arch_spin_unlock': /home/sof/src/arch/xtensa/include/arch/spinlock.h:121:11: error: variable 'result' set but not used [-Werror=unused-but-set-variable] 121 | uint32_t result; | ^~~~~~ [ 11%] No download step for 'rimage_ep' cc1: all warnings being treated as errors gmake[3]: *** [src/init/CMakeFiles/ext_manifest.dir/build.make:77: src/init/CMakeFiles/ext_manifest.dir/ext_manifest.c.o] Error 1 gmake[2]: *** [CMakeFiles/Makefile2:2118: src/init/CMakeFiles/ext_manifest.dir/all] Error 2 gmake[2]: *** Waiting for unfinished jobs.... [ 12%] Building ASM object src/arch/xtensa/hal/CMakeFiles/hal.dir/memcopy.S.o [ 13%] Building ASM object src/arch/xtensa/xtos/CMakeFiles/xtos.dir/core-restore.S.o [ 14%] No update step for 'smex_ep' [ 14%] Building ASM object src/arch/xtensa/xtos/CMakeFiles/xtos.dir/core-save.S.o

facing this error when building with GCC, not sure why this is not raised previously. Could that be due to GCC version?

Before getting into what you don't know, simply start by adding to the commit message what you know already: just the error message and the gcc version that prints it. That could be enough.

marc-hb · 2024-08-07T17:08:47Z

src/arch/xtensa/include/xtensa/core-macros.h

@@ -433,7 +433,7 @@ static inline unsigned  XTHAL_COMPARE_AND_SET( int *addr, int testval, int setva
 	    : "=a"(result) : "0" (setval), "a" (testval), "a" (addr)
 	    : "memory");
 #elif XCHAL_HAVE_INTERRUPTS
-    int tmp;
+    int tmp=0;


This fix is not ACP70-specific at all so it must be in a separate, standalone commit. The error message should be in the commit message.

marc-hb · 2024-08-07T17:14:44Z

src/arch/xtensa/include/xtensa/config/core.h

+#if defined(CONFIG_ACP_7_0)
+
+#ifndef UINT32_C
+#define UINT32_C(x) x


Please add a comment explaining why this is needed.

Do you really need BOTH #ifdef ACP70 AND #ifndef UINT32_C?

Yes, Actually ACP_7_0 platform uses the new 2023 xt-clang toolchain for XCC build. That need the defination of macro UINT32_C.

Just realized there is already something similar a few lines above. So we have some toolchain-specific ugliness with four #ifdef, none of which actually checks any toolchain stuff... This looks duplicated, messy and fragile. @paulstelian97 , @andyross, @dcpleung can you help maybe?

Just realized there is already something similar a few lines above. So we have some toolchain-specific ugliness with four #ifdef, none of which actually checks any toolchain stuff... This looks duplicated, messy and fragile. @paulstelian97 , @andyross, @dcpleung can you help maybe?

gentle reminder

IIRC, the old gcc-based xcc needs this. But newer xt-clang should be fine. Are you including stdint.h in correct place?

You can always do:

#if defined(__XT_CLANG__) #include <xtensa/xtensa-types.h> #endif

to limit it to xt-clang.

Condition #if defined(__XT_CLANG__) don't work( i.e., goes False) for GCC build. Both XCC and GCC builds require the inclusion of xtensa-types.h .

Including xtensa-types.h works. But need to add this file in the GCC build xtensa overlay tar balls(all old platforms) to avoid build errors. Otherwise need to include under conditional macro(CONFIG_ACP_7_0).

Here I wanted to convey that inclusion of xtensa-types.h in core.h file at line 59 as shown below works fine for both XCC and GCC builds. But had a concern of backward compatibility (build flow of existing platforms) i.e., this inclusion don't effect the existing platform's XCC build flow as all existing platform's Xtensa toolchains (have observed that in all existing AMD platforms) contain the header file xtensa-types.h, but effects the GCC build flow(i.e., cause build errors that xtensa-types.h is not found) of existing platform's (mostly all) tarball (toolchain) don't have file xtensa-types.h. Please note that inclusion of this header is required for both XCC and GCC builds of the platform ACP_7_0.

line 57 onwards in file core.h

/* CONFIGURATION INDEPENDENT DEFINITIONS: */ #ifdef __XTENSA__ #include <xtensa/hal.h> #include <xtensa/xtensa-types.h> #include <xtensa/xtensa-versions.h> #else #include "../hal.h" #include "../xtensa-versions.h" #endif

... or maybe just drop the __ZEPHYR__ guard. It's not clear why @paulstelian97 added it in #7538.

In any case this is a complex toolchain compatibility issue that definitely deserves a separate PR.

Created a separate PR for inclusion of xtensa-types.h file which has the macro UINT32_C.

GCC would not have __XT_CLANG__ declared so that should not affect the build... no?

I think we need clarification here. There are XCC (based on old GCC) and XT-CLANG (based on LLVM) compilers, and there is GCC which is not provided by Cadence. XCC build would not have __XT_CLANG__ declared either. Or are you saying that older XT-CLANG toolchain are having problems with including xtensa-types.h?

Thanks @dcpleung , can you please help transfer that question to newer, much smaller and focused #9413? It's hard enough to get attention to tricky issues like this, but it's much harder buried deep down this big, platform-specific PR named "ACP 7.0"

kv2019i

Only minor things, please check inline.

kv2019i · 2024-08-08T11:45:46Z

src/arch/xtensa/include/arch/spinlock.h

@@ -118,10 +118,7 @@ static inline void arch_spin_unlock(struct k_spinlock *lock)
 */


proper git summary prefix please

@saisurya-ch Can you update the git commit message to have proper prefix on the first line?

kv2019i · 2024-08-08T11:48:47Z

src/lib/alloc.c

@@ -92,6 +93,7 @@ static inline uint32_t heap_get_size(struct mm_heap *heap)

 	return size;
 }
+#endif


Can you remove L75-96? I assume clang complains that they are not used. It seems these have been dead code for long, not used for anything, so can be removed. Please add a separate commit.

As already discussed in commit bcbcec7 ("cmake: drop binutils-specific '-Wl,-EL' option") and commit ee58fef ("smex/cmake: move -Wl,EL option to target_linker_options() for clang") and their corresponding reviews on GitHub, this binutils-specific, endianness option makes no difference and is causing clang compatibility issues. It's now getting in the way of #9351 "Add the new platform ACP_7_0". I ran `./scripts/docker-run.sh ./scripts/xtensa-build-all.sh -a` with and without it and there was absolutely zero binary difference. I was added in 2019 by giant commit e0ba98d ("cmake: add CMakeLists for firmware build") without any rationale of why it would be needed. Signed-off-by: Marc Herbert <[email protected]>

This is required to allow building XTOS with newer, clang-based, Cadence toolchains as just submitted for ACP_7_0 in thesofproject#9351 Signed-off-by: Marc Herbert <[email protected]>

lyakh · 2024-08-12T14:33:13Z

src/platform/amd/acp_7_0/lib/clk.c

+			int_did_val = (uint32_t)(fraction_val * 100.0f);
+			fraction_val = (float)(int_did_val / 100.0f);
+
+			if (fraction_val == 0.0f)


I thought this wouldn't work, but after a brief test it does appear to work. Still, I'm not sure this would work always, maybe better do -1e-10 < x < 1e-10 or similar instead of == 0

Even if some rounding error "breaks" it for some values, similar rounding errors can affect x < 1e-10 too.

0.01 does not exist in float, it's more like: 0.00999999977648258209228515625
https://numeral-systems.com/ieee-754-converter/

In any case lines 227 and 228 are confusing and seem slower and unnecessary.

lyakh · 2024-08-12T14:36:08Z

src/platform/amd/acp_7_0/lib/clk.c

+{
+	volatile clk7_clk_pll_pwr_req_t clk7_clk_pll_pwr_req;
+	volatile clk7_clk_fsm_status_t  clk7_clk_fsm_status;
+	int count = 0;


superfluous init

lyakh · 2024-08-12T14:36:20Z

src/platform/amd/acp_7_0/lib/clk.c

+{
+	volatile clk7_clk_pll_pwr_req_t clk7_clk_pll_pwr_req;
+	volatile clk7_clk_fsm_status_t  clk7_clk_fsm_status;
+	int count = 0;


lyakh · 2024-08-12T14:39:24Z

src/platform/amd/acp_7_0/lib/clk.c

+
+	clk7_clk_pll_req_u_t  clk7_clk_pll_req;
+
+	clk7_clk_pll_req.u32all = 0;


clk7_clk_pll_req is completely overwritten in the next line, no need to set any members in it

This is required to allow building XTOS with newer, clang-based, Cadence toolchains as just submitted for ACP_7_0 in #9351 Signed-off-by: Marc Herbert <[email protected]>

stale review

marc-hb · 2024-08-14T16:58:29Z

I think the much smaller, non-AMD cleanups can be merged much faster if you submit them in a separate Pull Request

https://docs.zephyrproject.org/latest/contribute/contributor_expectations.html#defining-smaller-prs

scripts/set_xtensa_params.sh

src/platform/amd/acp_7_0/lib/clk.c

lyakh · 2024-08-15T06:59:19Z

src/platform/amd/acp_7_0/lib/clk.c

+
+void acp_7_0_reg_wait(void)
+{
+	int test_count = 0;


superfluous init

lyakh · 2024-08-15T07:01:26Z

src/platform/amd/acp_7_0/lib/clk.c

+	clk7_spll_field_2_t clk7_spll_field;
+	uint32_t spinediv      = 1;
+	float fract_part    = 0.0f;
+	float final_refclk  = 0.0f;


final_refclk is always calculated, no need to initialise

lyakh · 2024-08-15T07:03:27Z

src/platform/amd/acp_7_0/lib/clk.c

+	tr_info(&acp_clk_tr, "acp_change_clock_notify clock_freq : %d clock_type : %d",
+		clock_freq, clock_type);
+
+	fraction_val = (float)(clock_freq / (float)1000000.0f);


using f at the end already makes it a float constant, don't think the type-cast is needed

lyakh · 2024-08-15T07:04:23Z

src/platform/amd/acp_7_0/lib/clk.c

+		clock_freq, clock_type);
+
+	fraction_val = (float)(clock_freq / (float)1000000.0f);
+	clock_freq   = (clock_freq / 1000000);


superfluous parentheses and you can use /=

lyakh · 2024-08-15T07:06:18Z

src/platform/amd/acp_7_0/lib/clk.c

+	volatile clk7_clk1_dfs_status_u_t dfs_status;
+	volatile uint32_t updated_clk = 0;
+	float did = 0.0f;
+	float fraction_val = 0.0f;


superfluous init. Please check them everywhere

lyakh · 2024-08-15T07:09:01Z

src/platform/amd/acp_7_0/lib/clk.c

+		dfs_cntl.bitfields.CLK1_DIVIDER = 0x7F;
+		bypass_cntl.bitfields.CLK1_BYPASS_DIV = 0xF;
+	} else {
+		did  = (float)(boot_ref_clk / (float)fraction_val);


all variables are already of type float, no need for any type-casts

kv2019i

Last three commits need rewording of the git commit message.

kv2019i · 2024-08-28T09:06:55Z

src/arch/xtensa/include/arch/spinlock.h

@@ -118,10 +118,7 @@ static inline void arch_spin_unlock(struct k_spinlock *lock)
 */


@saisurya-ch Can you update the git commit message to have proper prefix on the first line?

saisurya-ch · 2024-08-28T09:32:12Z

Last three commits need rewording of the git commit message.

you mean the git commit prefixes such as feature, fix , refactor. right ?

kv2019i · 2024-08-28T16:55:32Z

Last three commits need rewording of the git commit message.

you mean the git commit prefixes such as feature, fix , refactor. right ?

Component/area prefix (same convention as in Linux and Zephyr). E.g. for src/arc/xtensa, here's example from recent git history:

work/sof$ git log --pretty=oneline src/arch/xtensa/ |head -20
ee5efdf5b9ef8996cebd2b95d159edef77c91656 src/arch/xtensa/CMakeLists.txt: drop bogus, non-portable -Wl,-EL
a2338e51035dfeb5eea520391486fa6d4aa3f2e0 rembrandt_defconfig: remove duplicate CONFIG_COMP_DCBLOCK=n
83f69f144fe8e1d770d78effdba971e46d27e4d7 mt8195: Disable GOOGLE AEC MOCK by default
a25c70e183f17e05bbb915c3c62a3ee97917491c amd: enable probe functionality config for AMD/Renoir platform.
a732b3bbd379797fff20485a64da7b4238db776f amd: add probe functionality in AMD/vangogh platform.
eb3fcc67b80cd7de6b71df81093c04dee61a2d5d amd: add probe functionality in AMD/rembrandt platform.
5cf1c9c0d67c09437bff0db6e3bda4c7a1b0fbb5 topology: Add Waves codec to MT8188 topology
2e5703d7d7b3438189285e7ca00898aa14caac59 amd: acp_6_3: add build support for ACP_6_3
2c4d02df1e952ad234a0ac2d006e152cba6048aa src: arch: xtensa: configs: add ACP_6_3 defconfig
2e7296f17f7082b7de77889e86335f529d90c948 cmake: add new ${RIMAGE_TOP} constant
54dd8ecf193732879f54786d88a0fe5cc89e02a3 arch: xtensa: remove tigerlake platform option
049329be0b52f102b16f04924664012441bcda81 arch: xtensa: configs: remove Intel Tiger Lake defconfigs
51556ddd63b0b7562716a3ba03a7a43b13ebffe5 audio: sof: remove -fno-inline-functions option for cavs2.5 and ace
b1683ec1d6efb6c922ea66c03668f965d4626740 amd: vangogh: add build support for vangogh
da39550a4f1304c99bc3f75fc90b08fa7dcabf50 src: arch: xtensa: configs: add defconfig
a8967f784cc18896513ce8a5002ca170b2971d69 platform: amd: create common platform code for different amd platforms
9d51b5c10f6011d12a61781113c38200f033c80e audio: delete switch component
138a9d018755ed04475f63a1a360c7af05eb8a82 smart_amp: modify the Kconfig hierarchy
c7af3c5b3be5bc374faafc9d3a4d598fce2d6efa Kconfig: don't fall back on CONFIG_TIGERLAKE
820fadae3930d1335bc22fda98633cc08c49b22d xtensa/CMakeLists.txt: simplify CONFIG_RENOIR/REMBRANDT logic

saisurya-ch · 2024-08-29T07:27:30Z

Last three commits need rewording of the git commit message.

you mean the git commit prefixes such as feature, fix , refactor. right ?

Component/area prefix (same convention as in Linux and Zephyr). E.g. for src/arc/xtensa, here's example from recent git history:

work/sof$ git log --pretty=oneline src/arch/xtensa/ |head -20
ee5efdf5b9ef8996cebd2b95d159edef77c91656 src/arch/xtensa/CMakeLists.txt: drop bogus, non-portable -Wl,-EL
a2338e51035dfeb5eea520391486fa6d4aa3f2e0 rembrandt_defconfig: remove duplicate CONFIG_COMP_DCBLOCK=n
83f69f144fe8e1d770d78effdba971e46d27e4d7 mt8195: Disable GOOGLE AEC MOCK by default
a25c70e183f17e05bbb915c3c62a3ee97917491c amd: enable probe functionality config for AMD/Renoir platform.
a732b3bbd379797fff20485a64da7b4238db776f amd: add probe functionality in AMD/vangogh platform.
eb3fcc67b80cd7de6b71df81093c04dee61a2d5d amd: add probe functionality in AMD/rembrandt platform.
5cf1c9c0d67c09437bff0db6e3bda4c7a1b0fbb5 topology: Add Waves codec to MT8188 topology
2e5703d7d7b3438189285e7ca00898aa14caac59 amd: acp_6_3: add build support for ACP_6_3
2c4d02df1e952ad234a0ac2d006e152cba6048aa src: arch: xtensa: configs: add ACP_6_3 defconfig
2e7296f17f7082b7de77889e86335f529d90c948 cmake: add new ${RIMAGE_TOP} constant
54dd8ecf193732879f54786d88a0fe5cc89e02a3 arch: xtensa: remove tigerlake platform option
049329be0b52f102b16f04924664012441bcda81 arch: xtensa: configs: remove Intel Tiger Lake defconfigs
51556ddd63b0b7562716a3ba03a7a43b13ebffe5 audio: sof: remove -fno-inline-functions option for cavs2.5 and ace
b1683ec1d6efb6c922ea66c03668f965d4626740 amd: vangogh: add build support for vangogh
da39550a4f1304c99bc3f75fc90b08fa7dcabf50 src: arch: xtensa: configs: add defconfig
a8967f784cc18896513ce8a5002ca170b2971d69 platform: amd: create common platform code for different amd platforms
9d51b5c10f6011d12a61781113c38200f033c80e audio: delete switch component
138a9d018755ed04475f63a1a360c7af05eb8a82 smart_amp: modify the Kconfig hierarchy
c7af3c5b3be5bc374faafc9d3a4d598fce2d6efa Kconfig: don't fall back on CONFIG_TIGERLAKE
820fadae3930d1335bc22fda98633cc08c49b22d xtensa/CMakeLists.txt: simplify CONFIG_RENOIR/REMBRANDT logic

Done

marc-hb · 2024-08-29T18:42:09Z

This depends on the resolution of #9413 -> #9442 first.

marc-hb · 2024-09-06T13:49:04Z

scripts/set_xtensa_params.sh

 	XTENSA_CORE="ACP_6_3_HiFi5_PROD_Linux"
 	HOST="xtensa-acp_6_3-elf"
 	TOOLCHAIN_VER="RI-2021.6-linux"
 	;;
+	acp_7_0)
+ 	PLATFORM="acp_7_0"


You can also remove PLATFORM now (and solve the small, new, corresponding git conflict)

Add support for ACP_7_0 in build scripts. Signed-off-by: SaiSurya Ch <[email protected]>

Add defconfig for ACP_7_0 platform. Signed-off-by: SaiSurya Ch <[email protected]>

Add ACP_7_0 platform topology. Signed-off-by: SaiSurya Ch <[email protected]>

Add acp_7_0 toml file to support sof-acp_7_0.ri binary build Signed-off-by: SaiSurya Ch <[email protected]>

Add build support to enable ACP_7_0 platform. Signed-off-by: SaiSurya Ch <[email protected]>

- Add ACP_7_0 platform support Signed-off-by: SaiSurya Ch <[email protected]>

Remove the unused variable 'result' that is causing the build error with GCC 11.4.0. error: variable 'result' set but not used [-Werror=unused-but-set-variable] Signed-off-by: SaiSurya Ch <[email protected]>

src: alloc: Remove unused get memory size functions. Signed-off-by: SaiSurya Ch <[email protected]>

amd: Remove inclusion of unrequired header files. Signed-off-by: SaiSurya Ch <[email protected]>

lgirdwood

@saisurya-ch fwiw, Intel has found that adding new platforms has been less effort after moving to Zephyr for SOF RTOS. Zephyr has a ton of platforms and a really refined process for adding new ones.

bhiregoudar · 2024-09-10T11:36:36Z

@saisurya-ch fwiw, Intel has found that adding new platforms has been less effort after moving to Zephyr for SOF RTOS. Zephyr has a ton of platforms and a really refined process for adding new ones.

Thanks Liam, we are in process of supporting Zephyr, soon we may get code up streamed.
Can we get this PR merged? we have some pending follow up patches in line.

saisurya-ch requested review from marc-hb, ranj063, bhiregoudar, sunilkumardommati, lgirdwood, plbossart, mmaka1, lbetlej, dbaluta and kv2019i as code owners August 7, 2024 09:32

This was referenced Aug 7, 2024

Eradicate -Wl,-EL for good #9353

Merged

Add mt8195 platform specific changes #4725

Merged

marc-hb previously requested changes Aug 7, 2024

View reviewed changes

kv2019i reviewed Aug 8, 2024

View reviewed changes

marc-hb mentioned this pull request Aug 9, 2024

Add $SOF_CC_BASE variable for building XTOS with clang-based, Cadence toolchains #9358

Merged

lyakh reviewed Aug 12, 2024

View reviewed changes

saisurya-ch closed this Aug 14, 2024

saisurya-ch force-pushed the main branch from e4e8a79 to 6c29ad3 Compare August 14, 2024 09:33

saisurya-ch reopened this Aug 14, 2024

marc-hb reviewed Aug 14, 2024

View reviewed changes

scripts/set_xtensa_params.sh Outdated Show resolved Hide resolved

lyakh reviewed Aug 15, 2024

View reviewed changes

saisurya-ch force-pushed the main branch 2 times, most recently from b1861b5 to 7165ac8 Compare August 21, 2024 06:50

kv2019i approved these changes Aug 26, 2024

View reviewed changes

bhiregoudar approved these changes Aug 26, 2024

View reviewed changes

marc-hb mentioned this pull request Aug 27, 2024

Fixes for i.MX xt-clang build #7538

Merged

saisurya-ch force-pushed the main branch from 7165ac8 to 2b7df53 Compare August 28, 2024 07:35

kv2019i reviewed Aug 28, 2024

View reviewed changes

marc-hb mentioned this pull request Aug 28, 2024

Include xtensa-types.h header file to support the ACP_7_0 platform's build. #9413

Closed

saisurya-ch force-pushed the main branch from 2b7df53 to e50055e Compare August 29, 2024 07:27

marc-hb marked this pull request as draft August 29, 2024 18:42

marc-hb marked this pull request as ready for review September 6, 2024 13:48

marc-hb reviewed Sep 6, 2024

View reviewed changes

saisurya-ch closed this Sep 8, 2024

saisurya-ch force-pushed the main branch from e50055e to 6c29ad3 Compare September 8, 2024 07:44

saisurya-ch added 9 commits September 8, 2024 13:29

scripts: add support for ACP_7_0 platform

884ca38

Add support for ACP_7_0 in build scripts. Signed-off-by: SaiSurya Ch <[email protected]>

src: arch: xtensa: configs: add ACP_7_0 defconfig

48e2ab1

Add defconfig for ACP_7_0 platform. Signed-off-by: SaiSurya Ch <[email protected]>

tools: topology: add ACP_7_0 topology support

d51c99b

Add ACP_7_0 platform topology. Signed-off-by: SaiSurya Ch <[email protected]>

rimage: config: add acp_7_0 toml file to support acp_7_0 build

620380b

Add acp_7_0 toml file to support sof-acp_7_0.ri binary build Signed-off-by: SaiSurya Ch <[email protected]>

amd: acp_7_0: add build support for ACP_7_0

93ad1cc

Add build support to enable ACP_7_0 platform. Signed-off-by: SaiSurya Ch <[email protected]>

platform: Add ACP_7_0 platform support

b5e0ccc

- Add ACP_7_0 platform support Signed-off-by: SaiSurya Ch <[email protected]>

spinlock: Remove unused variable 'result'

44b4649

Remove the unused variable 'result' that is causing the build error with GCC 11.4.0. error: variable 'result' set but not used [-Werror=unused-but-set-variable] Signed-off-by: SaiSurya Ch <[email protected]>

src: alloc: Remove unused functions

e64f20a

src: alloc: Remove unused get memory size functions. Signed-off-by: SaiSurya Ch <[email protected]>

amd: Remove inclusion of unrequired header files.

07f19d8

amd: Remove inclusion of unrequired header files. Signed-off-by: SaiSurya Ch <[email protected]>

saisurya-ch reopened this Sep 8, 2024

lgirdwood approved these changes Sep 9, 2024

View reviewed changes

kv2019i merged commit d00561e into thesofproject:main Sep 11, 2024
69 of 92 checks passed

		@@ -118,10 +118,7 @@ static inline void arch_spin_unlock(struct k_spinlock *lock)
		*/


		clk7_clk_pll_req_u_t clk7_clk_pll_req;

		clk7_clk_pll_req.u32all = 0;

Add the new platform ACP_7_0 #9351

Add the new platform ACP_7_0 #9351

Conversation

saisurya-ch commented Aug 7, 2024 • edited Loading

marc-hb left a comment • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

saisurya-ch Aug 9, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

marc-hb Aug 7, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

marc-hb Aug 7, 2024 • edited Loading

Choose a reason for hiding this comment

saisurya-ch Aug 9, 2024 • edited Loading

Choose a reason for hiding this comment

marc-hb Aug 9, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

saisurya-ch Aug 9, 2024 • edited Loading

Choose a reason for hiding this comment

marc-hb Aug 9, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

marc-hb Aug 19, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

saisurya-ch Aug 28, 2024 • edited Loading

Choose a reason for hiding this comment

saisurya-ch Aug 28, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kv2019i left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

marc-hb commented Aug 14, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kv2019i left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

saisurya-ch commented Aug 28, 2024 • edited Loading

kv2019i commented Aug 28, 2024

saisurya-ch commented Aug 29, 2024

marc-hb commented Aug 29, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

lgirdwood left a comment

Choose a reason for hiding this comment

bhiregoudar commented Sep 10, 2024

saisurya-ch commented Aug 7, 2024 •

edited

Loading

marc-hb left a comment •

edited

Loading

saisurya-ch Aug 9, 2024 •

edited

Loading

marc-hb Aug 7, 2024 •

edited

Loading

marc-hb Aug 7, 2024 •

edited

Loading

saisurya-ch Aug 9, 2024 •

edited

Loading

marc-hb Aug 9, 2024 •

edited

Loading

saisurya-ch Aug 9, 2024 •

edited

Loading

marc-hb Aug 9, 2024 •

edited

Loading

marc-hb Aug 19, 2024 •

edited

Loading

saisurya-ch Aug 28, 2024 •

edited

Loading

saisurya-ch Aug 28, 2024 •

edited

Loading

saisurya-ch commented Aug 28, 2024 •

edited

Loading

marc-hb commented Aug 29, 2024 •

edited

Loading