1）perf安装错误

图片.png如图，安装时找不到对应的包，尤其是linux-tools-5.4.0-107-generic

解决方法：

在docker内换源（改为阿里源之后安装成功）：

  1. $ sudo vim /etc/apt/sources.list
  7. deb http://mirrors.aliyun.com/ubuntu/ bionic main restricted universe multiverse
  9. deb http://mirrors.aliyun.com/ubuntu/ bionic-security main restricted universe multiverse
  11. deb http://mirrors.aliyun.com/ubuntu/ bionic-updates main restricted universe multiverse
  13. deb http://mirrors.aliyun.com/ubuntu/ bionic-proposed main restricted universe multiverse
  15. deb http://mirrors.aliyun.com/ubuntu/ bionic-backports main restricted universe multiverse
  17. deb-src http://mirrors.aliyun.com/ubuntu/ bionic main restricted universe multiverse
  19. deb-src http://mirrors.aliyun.com/ubuntu/ bionic-security main restricted universe multiverse
  21. deb-src http://mirrors.aliyun.com/ubuntu/ bionic-updates main restricted universe multiverse
  23. deb-src http://mirrors.aliyun.com/ubuntu/ bionic-proposed main restricted universe multiverse
  25. deb-src http://mirrors.aliyun.com/ubuntu/ bionic-backports main restricted universe multiverse

2）不同机器的perf.data读不出数据

perf record后的perf.data数据只能在本机上report并进行函数栈追踪，如果本机下载服务器的perf.data，将无法追踪函数栈，结果会全部都是planning_node，没有分析价值。
图片.png

解决方法：
过程：perf源码中有perf archive命令，用于将相关符号打包，方便在其它机器进行分析；但实际运用中报错perf archive is not a perf-command，可能是文档问题，故查看perf源码将perf-archive脚本从源码中提出，手动进行：

  1. #!/bin/bash
  2. # SPDX-License-Identifier: GPL-2.0
  3. # perf archive
  4. # Arnaldo Carvalho de Melo <acme@redhat.com>
  6. PERF_DATA=perf.data
  7. if [ $# -ne 0 ] ; then
  8.     PERF_DATA=$1
  9. fi
  11. #
  12. # PERF_BUILDID_DIR environment variable set by perf
  13. # path to buildid directory, default to $HOME/.debug
  14. #
  15. if [ -z $PERF_BUILDID_DIR ]; then
  16.     PERF_BUILDID_DIR=~/.debug/
  17. else
  18.         # append / to make substitutions work
  19.         PERF_BUILDID_DIR=$PERF_BUILDID_DIR/
  20. fi
  22. BUILDIDS=$(mktemp /tmp/perf-archive-buildids.XXXXXX)
  23. NOBUILDID=0000000000000000000000000000000000000000
  25. perf buildid-list -i $PERF_DATA --with-hits | grep -v "^$NOBUILDID " > $BUILDIDS
  26. if [ ! -s $BUILDIDS ] ; then
  27.     echo "perf archive: no build-ids found"
  28.     rm $BUILDIDS || true
  29.     exit 1
  30. fi
  32. MANIFEST=$(mktemp /tmp/perf-archive-manifest.XXXXXX)
  33. PERF_BUILDID_LINKDIR=$(readlink -f $PERF_BUILDID_DIR)/
  35. cut -d ' ' -f 1 $BUILDIDS | \
  36. while read build_id ; do
  37.     linkname=$PERF_BUILDID_DIR.build-id/${build_id:0:2}/${build_id:2}
  38.     filename=$(readlink -f $linkname)
  39.     echo ${linkname#$PERF_BUILDID_DIR} >> $MANIFEST
  40.     echo ${filename#$PERF_BUILDID_LINKDIR} >> $MANIFEST
  41. done
  43. tar cjf $PERF_DATA.tar.bz2 -C $PERF_BUILDID_DIR -T $MANIFEST
  44. rm $MANIFEST $BUILDIDS || true
  45. echo -e "Now please run:\n"
  46. echo -e "$ tar xvf $PERF_DATA.tar.bz2 -C ~/.debug\n"
  47. echo "wherever you need to run 'perf report' on."
  48. exit 0

在测试机上运行：
$ ./perf-archive.sh -i perf.data
生成符号表二进制文件：perf.data.tar.bz2

之后在任何机器上，只要复制perf.data和perf.data.tar.bz2就可以分析
运行命令：
$ tar xvf perf.data.tar.bz2 -C ~/.debug
图片.png图片.png