diff options
| author |   Michael Ellerman <mpe@ellerman.id.au> | 2023-08-01 20:26:50 +1000 |
|---|---|---|
| committer | Michael Ellerman <mpe@ellerman.id.au> | 2023-08-01 21:01:27 +1000 |
| commit | 2dc0bc1138eecc88b2c376ccb0b0acb215c25a5c (patch) | |
| tree | abf2ef33cd06a0cb680cac4bebd4d14757eae009 /arch/powerpc/kernel/head_64.S | |
| parent | powerpc/ftrace: Create a dummy stackframe to fix stack unwind (diff) | |
| download | wireguard-linux-2dc0bc1138eecc88b2c376ccb0b0acb215c25a5c.tar.xz wireguard-linux-2dc0bc1138eecc88b2c376ccb0b0acb215c25a5c.zip | |
powerpc/64e: Fix secondary thread bringup for ELFv2 kernels
When booting on e6500 with an ELF v2 ABI kernel, the secondary threads do
not start correctly:
[ 0.051118] smp: Bringing up secondary CPUs ...
[ 5.072700] Processor 1 is stuck.
This occurs because the startup code is written to use function
descriptors when loading the entry point for the secondary threads. When
building with ELF v2 ABI there are no function descriptors, and the code
loads junk values for the entry point address.
Fix it by using ppc_function_entry() in C, and DOTSYM() in asm, both of
which work correctly for ELF v2 ABI as well as ELF v1 ABI kernels.
Fixes: 8c5fa3b5c4df ("powerpc/64: Make ELFv2 the default for big-endian builds")
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20230801102650.48705-1-mpe@ellerman.id.au
Diffstat (limited to 'arch/powerpc/kernel/head_64.S')
| -rw-r--r-- | arch/powerpc/kernel/head_64.S | 3 |
1 files changed, 1 insertions, 2 deletions
diff --git a/arch/powerpc/kernel/head_64.S b/arch/powerpc/kernel/head_64.S index f132d8704263..6440b1bb332a 100644 --- a/arch/powerpc/kernel/head_64.S +++ b/arch/powerpc/kernel/head_64.S @@ -375,8 +375,7 @@ _GLOBAL(generic_secondary_smp_init) beq 20f /* start the specified thread */ - LOAD_REG_ADDR(r5, fsl_secondary_thread_init) - ld r4, 0(r5) + LOAD_REG_ADDR(r5, DOTSYM(fsl_secondary_thread_init)) bl book3e_start_thread /* stop the current thread */ |