Re: [Xen-devel] Xen S3 host resume: "Interrupts enabled after xen_acpi_processor_resume+0x0/0x34"

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On Tue, Nov 19, 2013 at 02:34:21PM -0500, Ben Guthro wrote:
> On Tue, Nov 19, 2013 at 12:57 PM, Jeremy Fitzhardinge <jeremy@goop.org>wrote:
> 
> > After an S3 host resume, I'm seeing:
> >
> > [48002.772629] ACPI: Low-level resume complete
> > [48002.772802] PM: Restoring platform NVS memory
> > [48002.776143] xen_acpi_processor: Uploading Xen processor PM info
> > [48002.776691] ------------[ cut here ]------------
> > [48002.776712] WARNING: CPU: 0 PID: 10235 at drivers/base/syscore.c:104
> > syscore_resume+0x9a/0xe0()
> > [48002.776751] Interrupts enabled after xen_acpi_processor_resume+0x0/0x34
> > [xen_acpi_processor]
> > [48002.776836] Modules linked in: tcp_lp btrfs zlib_deflate raid6_pq xor
> > ufs hfsplus hfs
> >  minix vfat msdos fat jfs xfs libcrc32c reiserfs xt_CHECKSUM tun hidp fuse
> > nf_conntrack_
> > netbios_ns nf_conntrack_broadcast ipt_MASQUERADE ip6t_REJECT xt_conntrack
> > ebtable_nat eb
> > table_broute bridge stp llc ebtable_filter ebtables ip6table_nat
> > nf_conntrack_ipv6 nf_de
> > frag_ipv6 nf_nat_ipv6 ip6table_mangle ip6table_security ip6table_raw
> > ip6table_filter ip6
> > _tables iptable_nat nf_conntrack_ipv4 nf_defrag_ipv4 nf_nat_ipv4 nf_nat
> > nf_conntrack ipt
> > able_mangle iptable_security iptable_raw rfcomm bnep snd_hda_codec_hdmi
> > snd_hda_codec_co
> > nexant iTCO_wdt iTCO_vendor_support arc4 x86_pkg_temp_thermal coretemp
> > iwldvm microcode
> > btusb snd_hda_intel mac80211 bluetooth joydev snd_hda_codec snd_hwdep
> > snd_seq serio_raw
> > iwlwifi i2c_i801 snd_seq_device
> > [48002.776894]  snd_pcm cfg80211 sdhci_pci sdhci mmc_core thinkpad_acpi
> > wmi tpm_tis e100
> > 0e rfkill snd_page_alloc tpm tpm_bios snd_timer ptp snd mei_me mei
> > soundcore shpchp lpc_
> > ich pps_core mfd_core mperf xen_acpi_processor xen_netback xen_blkback
> > xen_gntdev xen_ev
> > tchn xenfs xen_privcmd uinput binfmt_misc nfsd auth_rpcgss nfs_acl lockd
> > sunrpc dm_crypt
> >  crc32_pclmul crc32c_intel i915 ghash_clmulni_intel i2c_algo_bit
> > drm_kms_helper drm i2c_
> > core video
> > [48002.776903] CPU: 0 PID: 10235 Comm: systemd-sleep Not tainted
> > 3.11.7-200.fc19.x86_64 #1
> > [48002.776905] Hardware name: LENOVO 4286CTO/4286CTO, BIOS 8DET59WW (1.29
> > ) 04/02/2012
> > [48002.776917]  0000000000000009 ffff8801c7697d30 ffffffff8164777b
> > ffff8801c7697d78
> > [48002.776924]  ffff8801c7697d68 ffffffff8106715d ffffffffa01f6000
> > 0000000000000003
> > [48002.776931]  ffff8801c7697e08 ffff8801fd970000 0000000000000000
> > ffff8801c7697dc8
> > [48002.776933] Call Trace:
> > [48002.776949]  [<ffffffff8164777b>] dump_stack+0x45/0x56
> > [48002.776959]  [<ffffffff8106715d>] warn_slowpath_common+0x7d/0xa0
> > [48002.776972]  [<ffffffff810671cc>] warn_slowpath_fmt+0x4c/0x50
> > [48002.776989]  [<ffffffffa01f4bb0>] ?
> > xen_upload_processor_pm_data+0x300/0x300 [xen_acpi_processor]
> > [48002.776996]  [<ffffffff813f252a>] syscore_resume+0x9a/0xe0
> > [48002.777006]  [<ffffffff810abd7c>] suspend_devices_and_enter+0x40c/0x480
> > [48002.777014]  [<ffffffff810abf68>] pm_suspend+0x178/0x260
> > [48002.777020]  [<ffffffff810aae19>] state_store+0x79/0xf0
> > [48002.777033]  [<ffffffff812fc4af>] kobj_attr_store+0xf/0x20
> > [48002.777043]  [<ffffffff8121a226>] sysfs_write_file+0xc6/0x140
> > [48002.777054]  [<ffffffff811a819d>] vfs_write+0xbd/0x1e0
> > [48002.777062]  [<ffffffff811a8bd9>] SyS_write+0x49/0xa0
> > [48002.777073]  [<ffffffff81656999>] system_call_fastpath+0x16/0x1b
> > [48002.777077] ---[ end trace 15c5658c596be9f3 ]---
> > [48002.777081] Enabling non-boot CPUs ...
> >
> >
> > I wonder if this is a result of upload_pm_data()'s use of mutexes, which
> > being a sleeping synchronization operation, shouldn't be done in an
> > interrupt-off context.
> >
> > I'm also having the machine hang in the suspended state on resume
> > sometimes (that is, I open the lid/hit a key/press power to resume the
> > laptop, but the sleep light keeps pulsing and I can't get it out of that
> > state without a forced power-off. I wonder if these are related.
> >
> > This is with the stock Fedora 19 kernel, but I don't think there are any
> > differences from upstream in this area?
> >
> 
> There was a commit in 3.12 to handle systems with reduced hardware sleep -
> but that would be a failure 100% of the time - not intermittent.
> What version of Xen are you running?

I saw that on Xen 4.3 (or Xen 4.2) but when I upgraded to Xen 4.4 I didn't see
it anymore. Hence I was thinking it was Xen related.