MIPS: fix sibling call handling in get_frame_info
PATCH v3 1/2 MIPS: fix sibling call handling in get_frame_info
Given a function, get_frame_info() analyzes its instructions to figure out frame size and return address. get_frame_info() works as follows:
1. analyze up to 128 instructions if the function size is unknown
2. search for 'addiu/daddiu sp,sp,-immed' for frame size
3. search for 'sw ra,offset(sp)' for return address
4. end search when it sees jr/jal/jalr
This leads to an issue when the given function is a sibling call, example shown as follows.
801ca110 <schedule>:
801ca110: 8f820000 lw v0,0(gp) 801ca114: 8c420000 lw v0,0(v0) 801ca118: 080726f0 j 801c9bc0 <__schedule> 801ca11c: 00000000 nop
801ca120 <io_schedule>:
801ca120: 27bdffe8 addiu sp,sp,-24 801ca124: 3c028022 lui v0,0x8022 801ca128: afbf0014 sw ra,20(sp)
In this case, get_frame_info() cannot properly detect schedule's frame info, and eventually returns io_schedule's instead.
This patch adds 'j' to the end search condition to workaround sibling call cases.
--- arch/mips/kernel/process.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)
arch/mips/kernel/process.c +4 -2
@@ -231,8 +231,10 @@ static inline int is_ra_save_ins(union mips_instruction *ip) ip->i_format.rt ==31 ; } -static inline int is_jal_jalr_jr_ins (union mips_instruction *ip) +static inline int is_jump_ins (union mips_instruction *ip) { +if (ip->j_format.opcode == j_op) +return 1 ;if (ip->j_format.opcode == jal_op)return 1 ;if (ip->r_format.opcode != spec_op)
@@ -268,7 +270,7 @@ static int get_frame_info(struct mips_frame_info *info)for (i =0 ; i < max_insns; i++, ip++) { -if (is_jal_jalr_jr_ins (ip)) +if (is_jump_ins (ip))break ;if (!info->frame_size) {if (is_sp_move_ins (ip))
Signature · 1 line
1.7.10.2 (Apple Git-33)
Quoted 5 lines · Tony Wu
Date: Mon, 13 May 2013 00:04:29 +0800
From: Tony Wu <[email protected]>
To: [email protected], [email protected]
Subject: [PATCH v3 1/2] MIPS: fix sibling call handling inget_frame_info
Content-Type: text/plain; charset=us-ascii
I've applied both your patches but both did apply with fuzz so you may want to verify everything is working fine.
Thanks,
Ralf