lkml.org 
[lkml]   [2018]   [Nov]   [15]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Memory hotplug softlock issue
On 11/15/18 at 09:30am, Michal Hocko wrote:
> On Thu 15-11-18 15:53:56, Baoquan He wrote:
> > On 11/15/18 at 08:30am, Michal Hocko wrote:
> > > On Thu 15-11-18 13:10:34, Baoquan He wrote:
> > > > On 11/14/18 at 04:00pm, Michal Hocko wrote:
> > > > > On Wed 14-11-18 22:52:50, Baoquan He wrote:
> > > > > > On 11/14/18 at 10:01am, Michal Hocko wrote:
> > > > > > > I have seen an issue when the migration cannot make a forward progress
> > > > > > > because of a glibc page with a reference count bumping up and down. Most
> > > > > > > probable explanation is the faultaround code. I am working on this and
> > > > > > > will post a patch soon. In any case the migration should converge and if
> > > > > > > it doesn't do then there is a bug lurking somewhere.
> > > > > > >
> > > > > > > Failing on ENOMEM is a questionable thing. I haven't seen that happening
> > > > > > > wildly but if it is a case then I wouldn't be opposed.
> > > > > >
> > > > > > Applied your debugging patches, it helps a lot to printing message.
> > > > > >
> > > > > > Below is the dmesg log about the migrating failure. It can't pass
> > > > > > migrate_pages() and loop forever.
> > > > > >
> > > > > > [ +0.083841] migrating pfn 10fff7d0 failed
> > > > > > [ +0.000005] page:ffffea043ffdf400 count:208 mapcount:201 mapping:ffff888dff4bdda8 index:0x2
> > > > > > [ +0.012689] xfs_address_space_operations [xfs]
> > > > > > [ +0.000030] name:"stress"
> > > > > > [ +0.004556] flags: 0x5fffffc0000004(uptodate)
> > > > > > [ +0.007339] raw: 005fffffc0000004 ffffc900000e3d80 ffffc900000e3d80 ffff888dff4bdda8
> > > > > > [ +0.009488] raw: 0000000000000002 0000000000000000 000000cb000000c8 ffff888e7353d000
> > > > > > [ +0.007726] page->mem_cgroup:ffff888e7353d000
> > > > > > [ +0.084538] migrating pfn 10fff7d0 failed
> > > > > > [ +0.000006] page:ffffea043ffdf400 count:210 mapcount:201 mapping:ffff888dff4bdda8 index:0x2
> > > > > > [ +0.012798] xfs_address_space_operations [xfs]
> > > > > > [ +0.000034] name:"stress"
> > > > > > [ +0.004524] flags: 0x5fffffc0000004(uptodate)
> > > > > > [ +0.007068] raw: 005fffffc0000004 ffffc900000e3d80 ffffc900000e3d80 ffff888dff4bdda8
> > > > > > [ +0.009359] raw: 0000000000000002 0000000000000000 000000cb000000c8 ffff888e7353d000
> > > > > > [ +0.007728] page->mem_cgroup:ffff888e7353d000
> > > > >
> > > > > I wouldn't be surprised if this was a similar/same issue I've been
> > > > > chasing recently. Could you try to disable faultaround to see if that
> > > > > helps. It seems that it helped in my particular case but I am still
> > > > > waiting for the final good-to-go to post the patch as I do not own the
> > > > > workload which triggered that issue.
> > > >
> > > > Tried, still stuck in last block sometime. Usually after several times
> > > > of hotplug/unplug. If stop stress program, the last block will be
> > > > offlined immediately.
> > >
> > > Is the pattern still the same? I mean failing over few pages with
> > > reference count jumping up and down between attempts?
> >
> > ->count jumping up and down, mapcount stays the same value.
> >
> > >
> > > > [root@ ~]# cat /sys/kernel/debug/fault_around_bytes
> > > > 4096
> > >
> > > Can you make it 0?
> >
> > I executed 'echo 0 > fault_around_bytes', value less than one page size
> > will round up to one page.
>
> OK, I have missed that. So then there must be a different source of the
> page count volatility. Is it always the same file?
>
> I think we can rule out memory reclaim because that depends on the page
> lock. Is the stress test hitting on memory compaction? In other words,
> are
> grep compact /proc/vmstat
> counters changing during the offline test heavily? I am asking because I
> do not see compaction pfn walkers skipping over MIGRATE_ISOLATE
> pageblocks. But I might be missing something easily.
>
> It would be also good to find out whether this is fs specific. E.g. does
> it make any difference if you use a different one for your stress
> testing?

Created a ramdisk and put stress bin there, then run stress -m 200, now
seems it's stuck in libc-2.28.so migrating. And it's still xfs. So now xfs
is a big suspect. At bottom I paste numactl printing, you can see that it's
the last 4G.

Seems it's trying to migrate libc-2.28.so, but stress program keeps trying to
access and activate it.

[ 5055.461652] migrating pfn 190f4fb3e failed
[ 5055.461671] page:ffffea643d3ecf80 count:257 mapcount:251 mapping:ffff888e7a6ac528 index:0x85
[ 5055.474734] xfs_address_space_operations [xfs]
[ 5055.474742] name:"libc-2.28.so"
[ 5055.481070] flags: 0x1dfffffc0000026(referenced|uptodate|active)
[ 5055.490329] raw: 01dfffffc0000026 ffffc900000e3d80 ffffc900000e3d80 ffff888e7a6ac528
[ 5055.498080] raw: 0000000000000085 0000000000000000 000000fc000000f9 ffff88810a8f2000
[ 5055.505823] page->mem_cgroup:ffff88810a8f2000
[ 5056.335970] migrating pfn 190f4fb3e failed
[ 5056.335990] page:ffffea643d3ecf80 count:255 mapcount:250 mapping:ffff888e7a6ac528 index:0x85
[ 5056.348994] xfs_address_space_operations [xfs]
[ 5056.348998] name:"libc-2.28.so"
[ 5056.353555] flags: 0x1dfffffc0000026(referenced|uptodate|active)
[ 5056.364680] raw: 01dfffffc0000026 ffffc900000e3d80 ffffc900000e3d80 ffff888e7a6ac528
[ 5056.372428] raw: 0000000000000085 0000000000000000 000000fc000000f9 ffff88810a8f2000
[ 5056.380172] page->mem_cgroup:ffff88810a8f2000
[ 5057.332806] migrating pfn 190f4fb3e failed
[ 5057.332821] page:ffffea643d3ecf80 count:261 mapcount:250 mapping:ffff888e7a6ac528 index:0x85
[ 5057.345889] xfs_address_space_operations [xfs]
[ 5057.345900] name:"libc-2.28.so"
[ 5057.350451] flags: 0x1dfffffc0000026(referenced|uptodate|active)
[ 5057.359707] raw: 01dfffffc0000026 ffffc900000e3d80 ffffc900000e3d80 ffff888e7a6ac528
[ 5057.369285] raw: 0000000000000085 0000000000000000 000000fc000000f9 ffff88810a8f2000
[ 5057.377030] page->mem_cgroup:ffff88810a8f2000
[ 5058.285457] migrating pfn 190f4fb3e failed
[ 5058.285489] page:ffffea643d3ecf80 count:257 mapcount:250 mapping:ffff888e7a6ac528 index:0x85
[ 5058.298544] xfs_address_space_operations [xfs]
[ 5058.298556] name:"libc-2.28.so"
[ 5058.303092] flags: 0x1dfffffc0000026(referenced|uptodate|active)
[ 5058.314358] raw: 01dfffffc0000026 ffffc900000e3d80 ffffc900000e3d80 ffff888e7a6ac528
[ 5058.322109] raw: 0000000000000085 0000000000000000 000000fc000000f9 ffff88810a8f2000
[ 5058.329848] page->mem_cgroup:ffff88810a8f2000


[root@~]# numactl -H
available: 8 nodes (0-7)
node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161
node 0 size: 59817 MB
node 0 free: 54253 MB
node 1 cpus: 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179
node 1 size: 65536 MB
node 1 free: 61158 MB
node 2 cpus: 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197
node 2 size: 65536 MB
node 2 free: 62752 MB
node 3 cpus: 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215
node 3 size: 65536 MB
node 3 free: 62708 MB
node 4 cpus: 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233
node 4 size: 34816 MB
node 4 free: 24141 MB
node 5 cpus: 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251
node 5 size: 0 MB
node 5 free: 0 MB
node 6 cpus: 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269
node 6 size: 0 MB
node 6 free: 0 MB
node 7 cpus: 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287
node 7 size: 4096 MB
node 7 free: 6 MB
node distances:
node 0 1 2 3 4 5 6 7
0: 10 21 31 21 41 41 51 51
1: 21 10 21 31 41 41 51 51
2: 31 21 10 21 51 51 41 41
3: 21 31 21 10 51 51 41 41
4: 41 41 51 51 10 21 31 21
5: 41 41 51 51 21 10 21 31
6: 51 51 41 41 31 21 10 21
7: 51 51 41 41 21 31 21 10

\
 
 \ /
  Last update: 2018-11-15 14:12    [W:0.084 / U:0.760 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site