Mirrors/u-boot
2
0
Fork 0
mirror of https://github.com/AsahiLinux/u-boot synced 2024-12-17 00:33:06 +00:00
Code Issues Projects Releases Packages Wiki Activity
u-boot/arch/arm/cpu/armv8/fsl-layerscape/spintable.S

119 lines
2.8 KiB
ArmAsm
Raw Normal View History

armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
/* SPDX-License-Identifier: GPL-2.0+ */
/*
* (C) Copyright 2014-2015 Freescale Semiconductor
* Copyright 2019 NXP
*/
#include <config.h>
#include <linux/linkage.h>
#include <asm/macro.h>
#include <asm/system.h>
#include <asm/arch/mp.h>
armv8: layerscape: clean exported symbols in spintable.S Add a new variable secondary_boot_code_start, which holds a pointer to the start of the spin table code. This will help to relocate the code section. While at it, move the size variable from the end to the beginning so there is a common section for the variables. Remove any other symbols. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:34 +00:00
.align 3
.global secondary_boot_addr
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
secondary_boot_addr:
armv8: layerscape: clean exported symbols in spintable.S Add a new variable secondary_boot_code_start, which holds a pointer to the start of the spin table code. This will help to relocate the code section. While at it, move the size variable from the end to the beginning so there is a common section for the variables. Remove any other symbols. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:34 +00:00
.quad __secondary_boot_func
.global secondary_boot_code_start
secondary_boot_code_start:
.quad __secondary_boot_code_start
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
armv8: layerscape: clean exported symbols in spintable.S Add a new variable secondary_boot_code_start, which holds a pointer to the start of the spin table code. This will help to relocate the code section. While at it, move the size variable from the end to the beginning so there is a common section for the variables. Remove any other symbols. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:34 +00:00
.global secondary_boot_code_size
secondary_boot_code_size:
.quad __secondary_boot_code_end - __secondary_boot_code_start
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
/* Using 64 bit alignment since the spin table is accessed as data */
armv8: layerscape: fix alignment for spin table Fix the alignment so it will match the comments. The spin table has to be 8 byte aligned, so ".align 3" is enough. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:29 +00:00
.align 3
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
/* Secondary Boot Code starts here */
armv8: layerscape: clean exported symbols in spintable.S Add a new variable secondary_boot_code_start, which holds a pointer to the start of the spin table code. This will help to relocate the code section. While at it, move the size variable from the end to the beginning so there is a common section for the variables. Remove any other symbols. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:34 +00:00
__secondary_boot_code_start:
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
__spin_table:
.space CONFIG_MAX_CPUS*SPIN_TABLE_ELEM_SIZE
.align 2
armv8: layerscape: rework spin table There are two issues: (1) The spin table doesn't convert the endianness of the jump address. Although there is code for it, the result isn't used at all (x0). (2) If something goes wrong, the function returns. But that doesn't make sense at all. Use the actual converted jump address as destination to fix. If there is an error, jump to a trap loop. And rearrange the code exception level switching code to make it smaller and clearer. This reduces the size of the spin table code section from 696 bytes to 424 bytes. If CONFIG_ARMV8_SWITCH_TO_EL1 the code size reduced from 696 bytes to 632 bytes. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:36 +00:00
__secondary_boot_func:
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
/*
* MPIDR_EL1 Fields:
* MPIDR[1:0] = AFF0_CPUID <- Core ID (0,1)
* MPIDR[7:2] = AFF0_RES
* MPIDR[15:8] = AFF1_CLUSTERID <- Cluster ID (0,1,2,3)
* MPIDR[23:16] = AFF2_CLUSTERID
* MPIDR[24] = MT
* MPIDR[29:25] = RES0
* MPIDR[30] = U
* MPIDR[31] = ME
* MPIDR[39:32] = AFF3
*
* Linear Processor ID (LPID) calculation from MPIDR_EL1:
* (We only use AFF0_CPUID and AFF1_CLUSTERID for now
* until AFF2_CLUSTERID and AFF3 have non-zero values)
*
* LPID = MPIDR[15:8] | MPIDR[1:0]
*/
mrs x0, mpidr_el1
ubfm x1, x0, #8, #15
ubfm x2, x0, #0, #1
orr x10, x2, x1, lsl #2 /* x10 has LPID */
ubfm x9, x0, #0, #15 /* x9 contains MPIDR[15:0] */
/*
* offset of the spin table element for this core from start of spin
* table (each elem is padded to 64 bytes)
*/
lsl x1, x10, #6
armv8: layerscape: load function pointer using ADR Don't use LDR to load a pointer to a function. This will generate a literal which cannot be relocated. Use ADR which is PC-relative and therefore can easily be relocated. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:28 +00:00
adr x0, __spin_table
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
/* physical address of this cpus spin table element */
add x11, x1, x0
armv8: layerscape: load function pointer using ADR Don't use LDR to load a pointer to a function. This will generate a literal which cannot be relocated. Use ADR which is PC-relative and therefore can easily be relocated. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:28 +00:00
adr x0, __real_cntfrq
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
ldr x0, [x0]
msr cntfrq_el0, x0 /* set with real frequency */
str x9, [x11, #16] /* LPID */
mov x4, #1
str x4, [x11, #8] /* STATUS */
dsb sy
armv8: layerscape: rework spin table There are two issues: (1) The spin table doesn't convert the endianness of the jump address. Although there is code for it, the result isn't used at all (x0). (2) If something goes wrong, the function returns. But that doesn't make sense at all. Use the actual converted jump address as destination to fix. If there is an error, jump to a trap loop. And rearrange the code exception level switching code to make it smaller and clearer. This reduces the size of the spin table code section from 696 bytes to 424 bytes. If CONFIG_ARMV8_SWITCH_TO_EL1 the code size reduced from 696 bytes to 632 bytes. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:36 +00:00
1:
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
wfe
armv8: layerscape: rework spin table There are two issues: (1) The spin table doesn't convert the endianness of the jump address. Although there is code for it, the result isn't used at all (x0). (2) If something goes wrong, the function returns. But that doesn't make sense at all. Use the actual converted jump address as destination to fix. If there is an error, jump to a trap loop. And rearrange the code exception level switching code to make it smaller and clearer. This reduces the size of the spin table code section from 696 bytes to 424 bytes. If CONFIG_ARMV8_SWITCH_TO_EL1 the code size reduced from 696 bytes to 632 bytes. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:36 +00:00
ldr x4, [x11]
cbz x4, 1b
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
mrs x1, sctlr_el2
armv8: layerscape: rework spin table There are two issues: (1) The spin table doesn't convert the endianness of the jump address. Although there is code for it, the result isn't used at all (x0). (2) If something goes wrong, the function returns. But that doesn't make sense at all. Use the actual converted jump address as destination to fix. If there is an error, jump to a trap loop. And rearrange the code exception level switching code to make it smaller and clearer. This reduces the size of the spin table code section from 696 bytes to 424 bytes. If CONFIG_ARMV8_SWITCH_TO_EL1 the code size reduced from 696 bytes to 632 bytes. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:36 +00:00
tbz x1, #25, 2f
rev x4, x4 /* BE to LE conversion */
2:
ldr x6, =ES_TO_AARCH64
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
#ifdef CONFIG_ARMV8_SWITCH_TO_EL1
armv8: layerscape: rework spin table There are two issues: (1) The spin table doesn't convert the endianness of the jump address. Although there is code for it, the result isn't used at all (x0). (2) If something goes wrong, the function returns. But that doesn't make sense at all. Use the actual converted jump address as destination to fix. If there is an error, jump to a trap loop. And rearrange the code exception level switching code to make it smaller and clearer. This reduces the size of the spin table code section from 696 bytes to 424 bytes. If CONFIG_ARMV8_SWITCH_TO_EL1 the code size reduced from 696 bytes to 632 bytes. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:36 +00:00
adr x5, 3f
switch_el x7, 0f, _dead_loop, _dead_loop
0: armv8_switch_to_el2_m x5, x6, x7
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
#endif
armv8: layerscape: rework spin table There are two issues: (1) The spin table doesn't convert the endianness of the jump address. Although there is code for it, the result isn't used at all (x0). (2) If something goes wrong, the function returns. But that doesn't make sense at all. Use the actual converted jump address as destination to fix. If there is an error, jump to a trap loop. And rearrange the code exception level switching code to make it smaller and clearer. This reduces the size of the spin table code section from 696 bytes to 424 bytes. If CONFIG_ARMV8_SWITCH_TO_EL1 the code size reduced from 696 bytes to 632 bytes. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:36 +00:00
3:
ldr x7, [x11, #24] /* ARCH_COMP */
cbz x7, 4f
ldr x6, =ES_TO_AARCH32
4:
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
#ifdef CONFIG_ARMV8_SWITCH_TO_EL1
armv8: layerscape: rework spin table There are two issues: (1) The spin table doesn't convert the endianness of the jump address. Although there is code for it, the result isn't used at all (x0). (2) If something goes wrong, the function returns. But that doesn't make sense at all. Use the actual converted jump address as destination to fix. If there is an error, jump to a trap loop. And rearrange the code exception level switching code to make it smaller and clearer. This reduces the size of the spin table code section from 696 bytes to 424 bytes. If CONFIG_ARMV8_SWITCH_TO_EL1 the code size reduced from 696 bytes to 632 bytes. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:36 +00:00
switch_el x7, _dead_loop, 0f, _dead_loop
armv8: Disable pointer authentication traps for EL1 The use of ARMv8.3 pointer authentication (PAuth) is governed by fields in HCR_EL2, which trigger a 'trap to EL2' if not enabled. The reset value of these fields is 'architecturally unknown' so we must ensure that the fields are enabled (to disable the traps) if we are entering the kernel at EL1. The APK field disables PAuth instruction traps and the API field disables PAuth register traps Add code to disable the traps in armv8_switch_to_el1_m. Prior to doing so, it checks fields in the ID_AA64ISAR1_EL1 register to ensure pointer authentication is supported by the hardware. The runtime checks require a second temporary register, so add this to the EL1 transition macro signature and update 2 call sites. Signed-off-by: Peter Hoyes <Peter.Hoyes@arm.com>
2021-08-19 15:53:09 +00:00
0: armv8_switch_to_el1_m x4, x6, x7, x9
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
#else
armv8: layerscape: rework spin table There are two issues: (1) The spin table doesn't convert the endianness of the jump address. Although there is code for it, the result isn't used at all (x0). (2) If something goes wrong, the function returns. But that doesn't make sense at all. Use the actual converted jump address as destination to fix. If there is an error, jump to a trap loop. And rearrange the code exception level switching code to make it smaller and clearer. This reduces the size of the spin table code section from 696 bytes to 424 bytes. If CONFIG_ARMV8_SWITCH_TO_EL1 the code size reduced from 696 bytes to 632 bytes. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:36 +00:00
switch_el x7, 0f, _dead_loop, _dead_loop
0: armv8_switch_to_el2_m x4, x6, x7
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
#endif
armv8: layerscape: rework spin table There are two issues: (1) The spin table doesn't convert the endianness of the jump address. Although there is code for it, the result isn't used at all (x0). (2) If something goes wrong, the function returns. But that doesn't make sense at all. Use the actual converted jump address as destination to fix. If there is an error, jump to a trap loop. And rearrange the code exception level switching code to make it smaller and clearer. This reduces the size of the spin table code section from 696 bytes to 424 bytes. If CONFIG_ARMV8_SWITCH_TO_EL1 the code size reduced from 696 bytes to 632 bytes. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:36 +00:00
_dead_loop:
wfe
b _dead_loop
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
/* Ensure that the literals used by the secondary boot code are
* assembled within it (this is required so that we can protect
* this area with a single memreserve region
*/
.ltorg
/* 64 bit alignment for elements accessed as data */
armv8: layerscape: fix alignment for spin table Fix the alignment so it will match the comments. The spin table has to be 8 byte aligned, so ".align 3" is enough. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:29 +00:00
.align 3
armv8: layerscape: move spin table into own module Move it out of lowlevel.S into spintable.S. On layerscape, the secondary CPUs are brought up in main u-boot. This will make it possible to only compile the spin table code for the main u-boot and omit it in SPL. This saves about 720 bytes in the SPL. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:27 +00:00
.global __real_cntfrq
__real_cntfrq:
.quad COUNTER_FREQUENCY
/* Secondary Boot Code ends here */
armv8: layerscape: clean exported symbols in spintable.S Add a new variable secondary_boot_code_start, which holds a pointer to the start of the spin table code. This will help to relocate the code section. While at it, move the size variable from the end to the beginning so there is a common section for the variables. Remove any other symbols. Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>
2020-06-01 19:53:34 +00:00
__secondary_boot_code_end:
Reference in a new issue Copy permalink