|
|
EDA365欢迎您登录！
您需要 登录 才可以下载或查看，没有帐号？注册
x
最近在用海思的HI3515设计一款工控机，第一版做完后，开始开发第二版，主要改动是增删了部分硬件功能，DDR颗粒从原来的HY5PS1G1631CFP-Y5C换成了H5PS1G63EFRY5C。
; q0 w) b3 r* S: c0 ]( I: J Y 结果发现第二版的程序在引导过程中偶尔会出现Unable to handle kernel paging request的错误，且后边的地址经常会变化，并不是固定的错误。错误的引导信息如下：6 J7 z# ]! `* z8 d# k( a
Unable to handle kernel NULL pointer dereference at virtual address 0000003c% W, G, E" J8 X8 t' V. M6 z3 r
pgd = c1de4000
# _* k+ L" p6 c% R [0000003c] *pgd=c1de3031, *pte=00000000, *ppte=00000000
8 \8 J% @& n8 U0 }+ F9 | Internal error: Oops: 817 [#1]7 l, f. w* l: f$ L- E9 @: [7 `
Modules linked in:
2 o( A: q5 l7 x# T( N CPU: 0 Not tainted (2.6.24-rt1-hi3515v100 #24)" N: e+ Y; H( [. Q9 E3 {) U
PC is at generic_file_aio_read+0x20/0x1a4
3 z' ~) `$ H( d6 }; k3 _1 K m LR is at do_sync_read+0xc8/0x114
$ m5 ~% v( C; G' h6 B pc : [<c0060904>] lr : [<c007cc1c>] psr: a0000013" W" J" i8 o; X7 J7 z3 P2 t* ~
sp : c1dd9d18 ip : 00000000 fp : c1dd9d60( {: w6 e( A+ v0 R8 m6 O9 B
r10: 00000000 r9 : 00000000 r8 : c1dd9e506 Q" F1 o% `8 |
r7 : c1dd9d6c r6 : c1dbb660 r5 : c1dd9dbc r4 : c1dd9d74
% j; k* W. L/ |* ^8 @ r3 : 00000000 r2 : 00000001 r1 : 0000003c r0 : c1dd9d74
" g- r! N q6 a Flags: NzCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment kernel/ a& C: F. u! o$ U2 Y7 s
Control: 0005317f Table: c1de4000 DAC: 00000017: Y7 r6 y E8 X. G7 n
Process rcS (pid: 219, stack limit = 0xc1dd8258)
* J6 k7 ]6 @1 T/ g- G( \3 u Stack: (0xc1dd9d18 to 0xc1dda000)% H" E( R. \8 Z- k
9d00: c0093cc4 c00f95cc
; I5 U, A* d1 J6 `1 G5 [ 9d20: 00000000 c1dd9e28 c1dd9da4 c1dd9d38 c0085be8 c0093cb4 c1dd9d74 c1dd9dbc
+ I) K9 s5 q7 C, b3 n6 P5 U9 ]$ T 9d40: c1dbb660 c1dd9d6c c1dd9e50 00000000 00000000 c1dd9e24 c1dd9d68 c007cc1c
: Q- S- o$ S: j0 C& _ 9d60: c00608f8 00000000 00000000 c1d6bdc0 00000080 00000000 c1dd9fb0 000000008 @8 E8 [- @5 A
9d80: 00000001 ffffffff c1dbb660 00000000 00000000 00000000 00000000 c1c5c0e0# Z: X6 X* }) B" x3 y% F% d
9da0: 00000000 00000000 c1dd9de0 c1c5c0e0 c0049d48 c1dd9db4 c1dd9db4 000000004 \' x1 O; K2 t/ t% Y+ O
9dc0: 00000000 c0044454 c0044004 c1dd9df8 c1dbb660 c1dd9e08 00000080 c007bde0; o5 y$ j X2 e- z" T
9de0: c0065b7c c1dbb660 c1dd9e28 c1dd9fb0 c1dd8000 c0294ac8 c1dd9fb0 00000080
: N4 U \2 G( b% v7 ] 9e00: c1dbb660 c1d6bdc0 c1dd9e50 c007cb54 00000000 c1dd9fb0 c1dd9e4c c1dd9e28
. b6 G" Q# W3 p8 `# Z9 g- m! @ 9e20: c007cd14 c007cb64 c1801b40 bf000000 00000000 c1dd9e50 c1dd8000 c0294ac8
8 j0 G Q( T+ ]6 N. s5 v 9e40: c1dd9e70 c1dd9e50 c00814b8 c007cc78 00000000 00000000 00000080 c1d6bdc0
* E/ o: @ m" K 9e60: 00000000 c1dd9e8c c1dd9e74 c0081d24 c0081474 c1dd9e98 c1d6bdc0 c1dd9fb0
$ T6 {' {0 U3 m: W- a; l8 Y 9e80: c1dd9f30 c1dd9e90 c00ad608 c0081c98 c1d6bdc2 00000000 6e69622f 0068732f
$ M) Q* z {1 n/ \ 9ea0: c1db2320 beffff7b c1dd9ee8 c1dd9eb8 c006e3e4 00000000 00000020 00000000$ u& _$ J# w1 j3 ~2 o. p- m
9ec0: c1dd8000 00000001 00000000 c1d6bdc0 00000000 00000000 00000000 c1dd9f1c
# [5 s! Z9 X: @ 9ee0: 00000000 00000000 00000000 c1dd9fb0 00000000 c1dd9efc 00000000 00000f7b
8 e; [/ x/ [% ^; o8 I0 i- N' M 9f00: c1dd9f1c c1dd9f10 c0294e54 c1d6bdc0 fffffffe c1dd8000 c0294e34 c1d6bdc02 ?- y* G# x2 n# X) v9 M4 W
9f20: fffffff8 c1dd9f5c c1dd9f34 c0081fc0 c00ad3fc c1d6bdc0 beffff92 00000000: Q: ~7 n& a3 r0 w ^& k( @
9f40: 001d7bcc 001d7bc4 c1dd8000 c1dd9fb0 c1dd9f84 c1dd9f60 c0082248 c0081f2c+ I& g5 w) k, l- k" f! h# Q1 N
9f60: c1c99000 001d7bcc c1dd9fb0 c1c99000 c001ffe4 001c71fc c1dd9fa4 c1dd9f88+ V7 m( b' W5 L$ b
9f80: c00237c4 c0082124 001d7bc4 001d7ba4 00000001 0000000b 00000000 c1dd9fa8
' [: I, J2 {# p9 u 9fa0: c001fe40 c0023798 001d7bc4 001d7ba4 001d7ba4 001d7bc4 001d7bcc 001d7ba4& C% `8 k7 Y m; {2 v3 N- ?
9fc0: 001d7bc4 001d7ba4 00000001 001d7bcc 001d7bcc 00000001 001c71fc 00000000" H* T# s2 V0 j- |$ R8 }
9fe0: 400833c4 be813a44 00069d44 400833cc 20000010 001d7ba4 00000000 00000000, n5 D; ?/ v' Q9 n, w {
Backtrace:* {2 c n' k5 v. |# R
[<c00608e8>] (generic_file_aio_read+0x4/0x1a4) from [<c007cc1c>] (do_sync_read+0xc8/0x114)
! i0 E% b/ v7 f$ W: h [<c007cb54>] (do_sync_read+0x0/0x114) from [<c007cd14>] (vfs_read+0xac/0x144): K1 Q/ q& r. V! c8 W. |2 q, r7 y
[<c007cc68>] (vfs_read+0x0/0x144) from [<c00814b8>] (kernel_read+0x54/0x84)
4 Q. P8 n' p h$ x# F2 H r8:c0294ac8 r7:c1dd8000 r6:c1dd9e50 r5:00000000 r4:bf000000: X& h. M% e3 N7 P$ g! c, l
[<c0081464>] (kernel_read+0x0/0x84) from [<c0081d24>] (prepare_binprm+0x9c/0x108)
) \: x& a! V+ z8 A r6:00000000 r5:c1d6bdc0 r4:00000080) @7 s9 L0 i5 s6 Y% I
[<c0081c88>] (prepare_binprm+0x0/0x108) from [<c00ad608>] (load_script+0x21c/0x240)
. _& ?2 f1 |" e4 H r6:c1dd9fb0 r5:c1d6bdc0 r4:c1dd9e98$ e' k& }/ E8 A3 {! c
[<c00ad3ec>] (load_script+0x0/0x240) from [<c0081fc0>] (search_binary_handler+0xa4/0x1f8)) G! U6 Y& \7 S& m* c" o4 ^
r6:fffffff8 r5:c1d6bdc0 r4:c0294e34
) x4 Y4 E" t2 F! [. }7 R/ T [<c0081f1c>] (search_binary_handler+0x0/0x1f8) from [<c0082248>] (do_execve+0x134/0x184)
/ G- g/ F; d1 ]6 F7 o [<c0082114>] (do_execve+0x0/0x184) from [<c00237c4>] (sys_execve+0x3c/0x5c)7 I) o2 T9 B: h1 T( d3 `
[<c0023788>] (sys_execve+0x0/0x5c) from [<c001fe40>] (ret_fast_syscall+0x0/0x2c)7 E5 w Q9 e. J2 a# |
r7:0000000b r6:00000001 r5:001d7ba4 r4:001d7bc4
: m7 [2 B: y/ z9 k1 G Code: e24dd020 e3a0c000 e58b3004 e1a04000 (e1a0a001)6 c, y' m- j) w
---[ end trace e292f16b7bf51848 ]---
( ?( r6 `& M3 q" D- }# E) v Segmentation fault
! ?2 T3 N# p8 Y1 y F6 N: {我先是把第一版和第二版上的NAND FLASH颗粒进行互换，发现新旧板上的颗粒在第一版的板子上都没有报错，而两个颗粒在第二版的板子上都会出现报错，初步确定问题是随着板子走的，而不是颗粒。
/ x: r7 D R, F4 ^0 Y0 H% | 然后将第二版上的硬件改动还原，尽量还原到和第一版一样，然后启动还是会出现报错信息，说明外围硬件的改动没有影响到引导驱动程序出错。
$ l; G4 x3 I M 现在在把新板子寄回工厂，准备把DDR颗粒换成和第一版一样，然后再尝试一下，这个需要一些时间。, t% H2 p. M7 L" ]# P3 c2 Z
然后比对了前后两种DDR颗粒的spec，未见明显的参数差别。
! R& s g- j0 l
+ ]# v9 j' H( Z& Y& v) ~8 z" T* `% P- X; z
PS:
) l2 r* \1 h/ P8 W这几天集中研究了一下这个问题，汇报一下：
; B% ?, z4 R2 G 1.启动UBOOT后，用网络下载后边的文件到内存中的方式来启动，问题依旧会出现，排除掉NAND flash的问题。
3 S: U3 H+ X2 [" j, }! o7 W% F 2.抓了1.8V，VREF_DDR的电压波形,没有发现drop和ripple.排除电压的问题0 c& G6 w+ y7 t7 e2 Z) O& D4 g
3.把老板子上HY5PS1G1631CLFP-Y5C和新板子上的H5PS1G63EFR-Y5C互换了一下。旧版运行到现在一次都没出错2 V% b2 |" l% `
而新版运行到现在报过一次错。) [& n" o6 L3 h! x* G
4.买了几颗HY5PS1G1631CFP-Y5C（与原来板子上DDR比较少了一个L,原来的DDR颗粒停产了，现在这个颗粒IDD6电流略高，其他参数一样）换到板上，暂时还没看到报错，还需多跑跑看。
) i8 u* O9 V( e( ?) {: D$ ^ 5.基于以上的实验，感觉是layout的问题，查了新旧版的走线，是完全一样的，etch length和manhattan length一样。
' E* g1 R: w: H! ?6 O 6.准备再次比对堆叠厚度，因为前后两版PCB是不同的厂子压合的，板子堆叠厚度是根据我提供的匹配阻抗由板厂自己计算的。 |
|
/1 
发表于 2013-2-1 12:39
收藏
淘帖
支持！
反对！