LoongArch: Jump to the link address before enable PG

The kernel entry points of both boot CPU (i.e., kernel_entry) and non-
boot CPUs (i.e., smpboot_entry) may be physical address from BootLoader
(in DA mode or identity-mapping PG mode). So we should jump to the link
address before PG enabled (because DA is disabled at the same time) and
just after DMW configured.

Specifically: With some older firmwares, non-boot CPUs started with PG
enabled, but this need firmware cooperation in the form of a temporary
page table, which is deemed unnecessary. OTOH, latest firmware versions
configure the non-boot CPUs to start in DA mode, so kernel-side changes
are needed.

Reviewed-by: WANG Xuerui <git@xen0n.name>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
This commit is contained in:
Huacai Chen 2022-08-11 20:52:12 +08:00
parent 5e8be07ca8
commit ab2579d795

View File

@ -21,6 +21,12 @@ SYM_CODE_START(kernel_entry) # kernel entry point
csrwr t0, LOONGARCH_CSR_DMWIN0
li.d t0, CSR_DMW1_INIT # CA, PLV0, 0x9000 xxxx xxxx xxxx
csrwr t0, LOONGARCH_CSR_DMWIN1
/* We might not get launched at the address the kernel is linked to,
so we jump there. */
la.abs t0, 0f
jr t0
0:
/* Enable PG */
li.w t0, 0xb0 # PLV=0, IE=0, PG=1
csrwr t0, LOONGARCH_CSR_CRMD
@ -29,11 +35,6 @@ SYM_CODE_START(kernel_entry) # kernel entry point
li.w t0, 0x00 # FPE=0, SXE=0, ASXE=0, BTE=0
csrwr t0, LOONGARCH_CSR_EUEN
/* We might not get launched at the address the kernel is linked to,
so we jump there. */
la.abs t0, 0f
jr t0
0:
la t0, __bss_start # clear .bss
st.d zero, t0, 0
la t1, __bss_stop - LONGSIZE
@ -74,6 +75,11 @@ SYM_CODE_START(smpboot_entry)
csrwr t0, LOONGARCH_CSR_DMWIN0
li.d t0, CSR_DMW1_INIT # CA, PLV0
csrwr t0, LOONGARCH_CSR_DMWIN1
la.abs t0, 0f
jr t0
0:
/* Enable PG */
li.w t0, 0xb0 # PLV=0, IE=0, PG=1
csrwr t0, LOONGARCH_CSR_CRMD
li.w t0, 0x04 # PLV=0, PIE=1, PWE=0
@ -85,9 +91,6 @@ SYM_CODE_START(smpboot_entry)
ld.d sp, t0, CPU_BOOT_STACK
ld.d tp, t0, CPU_BOOT_TINFO
la.abs t0, 0f
jr t0
0:
bl start_secondary
SYM_CODE_END(smpboot_entry)