Running an fio (fio-2.1.10) job against a software RAID-10 device built from multipathed SSDs causes all memory to be consumed by Acpi-State objects. This is the report from slabtop: Active / Total Objects (% used) : -1033565817 / -1033409379 (100.0%) Active / Total Slabs (% used) : 65344141 / 65345079 (100.0%) Active / Total Caches (% used) : 69 / 122 (56.6%) Active / Total Size (% used) : 256421282.18K / 256452679.19K (100.0%) Minimum / Average / Maximum Object : 0.02K / 0.08K / 4096.00K OBJS ACTIVE USE OBJ SIZE SLABS OBJ/SLAB CACHE SIZE NAME 3260693700 3260690441 1% 0.08K 65213874 50 260855496K Acpi-State 193936 163748 84% 0.03K 1564 124 6256K kmalloc-32 This is the script running the fio jobs: DEV='/dev/md6'=~ [yY](es)* ]] DRIVESIZEBYTES=$(($(cat /sys/block/`basename $DEV`/size) *512)) DRIVESIZEGiB=$(($DRIVESIZEBYTES/1024/1024/1024))e drive fill x 2" FIO="sudo /opt/clearsky/bin/fio" echo Drive Size=$DRIVESIZEGiB GiB $FIO --readwrite=rw --bs=4k --invalidate=1 --bsrange=4k-512k\ trap "exit 1;" SIGHUP SIGINT SIGTERM --direct=1 --size="$DRIVESIZEGiB"G\ --group_reporting --ioengine=libaio --numjobs=1\ devs=` ls /dev/md[345] `-minimal $names >> '/tmp/seq_precon_log.txt'; names=""Pass # $n, complete drive fill" done for dev in $devs; do names=`echo $names "--name=${dev} --iodepth=128 --filename=${dev}"`; doneit fi echo Conditioning $devs read -p "Are you sure you want to continue? <y/N> " prompt ~ if [[ $prompt =~ [yY](es)* ]] then echo "Starting 512KB sequential write, complete drive fill x 2" date n=0; while (( n++ < 2000 )); do $FIO --readwrite=rw --bs=4k --invalidate=1 --bsrange=4k-512k\ --rwmixread=33 --end_fsync=0 --direct=1 --size="$DRIVESIZEGiB"G\ --group_reporting --ioengine=libaio --numjobs=1\ --randrepeat=0 --minimal $names >> '/tmp/seq_precon_log.txt'; echo "Pass # $n, complete drive fill" done date else exit fi
If --ioengine=libaio is removed the issue goes away. This is the slabtop from the run without ioengine: Active / Total Objects (% used) : 1172329 / 1264471 (92.7%) Active / Total Slabs (% used) : 147877 / 148087 (99.9%) Active / Total Caches (% used) : 72 / 113 (63.7%) Active / Total Size (% used) : 1770060.91K / 1782121.35K (99.3%) Minimum / Average / Maximum Object : 0.02K / 1.41K / 4096.00K OBJS ACTIVE USE OBJ SIZE SLABS OBJ/SLAB CACHE SIZE NAME 274833 269544 98% 0.10K 7047 39 28188K buffer_head 203980 168866 82% 0.03K 1645 124 6580K kmalloc-32 121248 112230 92% 0.25K 7578 16 30312K kmalloc-256 112728 111322 98% 0.07K 2013 56 8052K Acpi-ParseExt 97544 97521 99% 16.00K 97544 1 1560704K kmalloc-16384 78435 77473 98% 0.19K 3735 21 14940K dentry 59829 58472 97% 0.12K 1813 33 7252K kernfs_node_cache 52794 43897 83% 0.06K 838 63 3352K kmalloc-64 30250 28222 93% 0.08K 605 50 2420K Acpi-State 26922 26894 99% 0.53K 3846 7 15384K inode_cache 22785 22366 98% 0.12K 735 31 2940K kmalloc-96 22572 16167 71% 0.04K 228 99 912K Acpi-Namespace 17556 16389 93% 0.19K 836 21 3344K kmalloc-192 15344 15193 99% 0.50K 1918 8 7672K kmalloc-512 Adding --ioengine=libaio in again shows Acpi-State slab growing without bound. I was also unable to reproduce this with a similar vdbench script: *workload: Single run, 3 raw disks, 0% random, 33% read of 4k blocks at unlimited rate *HD: Host Definition *SD: Storage Definition *WD: Workload Definition *RD: Run Definition * hd=default,user=root,shell=ssh,jvms=8 sd=sd1,lun=/dev/md3,openflags=o_direct,hitarea=0,range=(0,100),threads=8 sd=sd2,lun=/dev/md4,openflags=o_direct,hitarea=0,range=(0,100),threads=8 sd=sd3,lun=/dev/md5,openflags=o_direct,hitarea=0,range=(0,100),threads=8 wd=wd1,sd=(sd1,sd2,sd3),xfersize=(4096,100),rdpct=33,seekpct=0 rd=run1,wd=wd1,iorate=max,elapsed=1800,warmup=600
Please remove the "'=~ [yY](es)* ]] from the first line of the script in the original comment. It is a cut and paste error.
I upgraded fio: # ./fio --version fio-2.12 using ioengine=libaio, the results are the same: Active / Total Objects (% used) : 170404994 / 170852831 (99.7%) Active / Total Slabs (% used) : 3574679 / 3574916 (100.0%) Active / Total Caches (% used) : 70 / 122 (57.4%) Active / Total Size (% used) : 15112939.94K / 15169514.47K (99.6%) Minimum / Average / Maximum Object : 0.02K / 0.09K / 4096.00K OBJS ACTIVE USE OBJ SIZE SLABS OBJ/SLAB CACHE SIZE NAME 168280800 168280800 23% 0.08K 3365616 50 13462464K Acpi-State 645918 444330 68% 0.10K 16562 39 66248K buffer_head 324508 200318 61% 0.03K 2617 124 10468K kmalloc-32 323296 268628 83% 0.25K 20206 16 80824K kmalloc-256 254331 242560 95% 0.19K 12111 21 48444K dentry 165528 158701 95% 0.04K 1672 99 6688K Acpi-Namespace 117306 111044 94% 0.06K 1862 63 7448K kmalloc-64 115008 115008 100% 0.31K 9584 12 38336K bio-3 112840 111079 98% 0.07K 2015 56 8060K Acpi-ParseExt
I eliminated both the MD RAID-10 and the device mapper multi-path to run the test on the raw devices (/dev/sd?) with the same defect: Active / Total Caches (% used) : 72 / 112 (64.3%) Active / Total Size (% used) : 149415419.76K / 149439781.30K (100.0%) Minimum / Average / Maximum Object : 0.02K / 0.08K / 4096.00K OBJS ACTIVE USE OBJ SIZE SLABS OBJ/SLAB CACHE SIZE NAME 1890042650 1890042295 0% 0.08K 37800853 50 151203412K Acpi-State 193869 187373 96% 0.10K 4971 39 19884K buffer_head 257052 167477 65% 0.03K 2073 124 8292K kmalloc-32 152192 115624 75% 0.25K 9512 16 38048K kmalloc-256 112896 111877 99% 0.07K 2016 56 8064K Acpi-ParseExt 97565 97489 99% 16.00K 97565 1 1561040K kmalloc-16384 85932 81944 95% 0.19K 4092 21 16368K dentry 57288 53843 93% 0.12K 1736 33 6944K kernfs_node_cache 48636 42996 88% 0.06K 772 63 3088K kmalloc-64 40572 40403 99% 0.53K 5796 7 23184K inode_cache 22847 21776 95% 0.12K 737 31 2948K kmalloc-96 15225 13729 90% 0.19K 725 21 2900K kmalloc-192 15939 13178 82% 0.04K 161 99 644K Acpi-Namespace 13008 12336 94% 0.50K 1626 8 6504K kmalloc-5
Correction to comment 5: MD RAID-10 was eliminated. Multi-path was not. The defect does not appear to be present on the raw devices. The following is the slabtop output after twice as much I/O as shown in comment 5: Active / Total Objects (% used) : 1293649 / 1470729 (88.0%) Active / Total Slabs (% used) : 148968 / 149123 (99.9%) Active / Total Caches (% used) : 67 / 110 (60.9%) Active / Total Size (% used) : 1756033.81K / 1785582.46K (98.3%) Minimum / Average / Maximum Object : 0.02K / 1.21K / 4096.00K OBJS ACTIVE USE OBJ SIZE SLABS OBJ/SLAB CACHE SIZE NAME 270777 266887 98% 0.10K 6943 39 27772K buffer_head 235200 234330 99% 0.08K 4704 50 18816K Acpi-State 232252 164813 70% 0.03K 1873 124 7492K kmalloc-32 112952 110903 98% 0.07K 2017 56 8068K Acpi-ParseExt 163200 109900 67% 0.25K 10200 16 40800K kmalloc-256 97555 97518 99% 16.00K 97555 1 1560880K kmalloc-16384 69174 61546 88% 0.19K 3294 21 13176K dentry 55143 53400 96% 0.12K 1671 33 6684K kernfs_node_cache 48888 40602 83% 0.06K 776 63 3104K kmalloc-64
I have demonstrated that the defect is only on the write path.
Please post the acpidump for this machine, thanks.
Created attachment 224461 [details] requested acpidump file. As requested.
Thanks. Please also attach your fio script as a file. Even after handling the copy/paste error I cannot run the script and don't want to assume any corrections.
Created attachment 224471 [details] fio script As requested.
dmesg would also help. thanks.
Created attachment 225271 [details] system dmesg output As requested
I booted linux 4.6.0 and the fault is not present.
I downloaded and built 4.4.15, the fault did not reproduce.
It would be helpful to us to know the exact platform where this was seen. thanks.
Created attachment 225531 [details] acpi state debug patch This patch will print a number that corresponds to the Acpi-State descriptor being created. Types are enumerated in drivers/acpi/acpica/acobject.h. Look for the prefix ACPI_DESC_TYPE.
Hi Matthew. If you can move to a newer kernel that would be best. If you can't or would like to pursue the root cause if this problem please try the above patch on the problem kernel to see which descriptor is behind the memory problem. Should print to dmesg. You may want to clear dmesg (save it first if desired) before running your script.
I get the following compilation errors when building linux 4.1.8 with your patch: In file included from include/acpi/acpi.h:61:0, from drivers/acpi/acpica/utstate.c:44: drivers/acpi/acpica/utstate.c: In function ‘acpi_ut_create_generic_state’: include/acpi/acoutput.h:213:41: error: too few arguments to function ‘acpi_info’ #define ACPI_INFO(plist) acpi_info plist ^ drivers/acpi/acpica/utstate.c:133:2: note: in expansion of macro ‘ACPI_INFO’ ACPI_INFO(("Acpi Generic State Created, Type: 0x%x\n", ^ In file included from include/acpi/acpi.h:64:0, from drivers/acpi/acpica/utstate.c:44: include/acpi/acpixf.h:868:5: note: declared here acpi_info(const char *module_name, ^ include/acpi/acpixf.h:320:2: note: in definition of macro ‘ACPI_MSG_DEPENDENT_RETURN_VOID’ prototype; ^ scripts/Makefile.build:258: recipe for target 'drivers/acpi/acpica/utstate.o' failed make[3]: *** [drivers/acpi/acpica/utstate.o] Error 1 make[3]: *** Waiting for unfinished jobs.... scripts/Makefile.build:403: recipe for target 'drivers/acpi/acpica' failed make[2]: *** [drivers/acpi/acpica] Error 2 scripts/Makefile.build:403: recipe for target 'drivers/acpi' failed make[1]: *** [drivers/acpi] Error 2 Makefile:947: recipe for target 'drivers' failed make: *** [drivers] Error 2 We are not in a position to move forward to a new kernel at this time. We have not yet targeted our next kernel. We may want to wait until the next new long term support kernel is available.
Created attachment 226361 [details] debug patch for 4.1 kernel My apologies. It applies to the current kernel. Simple fix. Compiled tested on 4.1.8.
Here is the BOM for the systems in question: Description Quantity OEM Dell Storage MD1420, 24 HDs 2.5", 2Us (210-ADBU) 1 OEM SHIP,MD1420,DAO (340-AKRE) 1 No Raid, no Controller (405-AAFC) 1 Brand/No-Bezel, 2U, OEM Storage (325-BBIQ) 1 Enclosure Management Module, dual (403-BBGM) 1 Hard Drive Filler 2.5in, single blank (400-AEPR) 10 200GB Solid State Drive SAS Write Intensive MLC 12Gbps 2.5in Hot-plug Drive, PX04SH (400-AMEY) 14 Rack rail, 2Us, Static (770-BBJE) 1 12Gb HD-Mini to HD-Mini SAS cable, 0.5m (470-ABDN) 4 Solutions Program Support (927-3179) 1 Description Quantity OEM PowerEdge R630 (210-ADCY) 1 PowerEdge R630 Motherboard MLK (329-BCZI) 1 OEM PowerEdge R630 Shipping-8 Drive Chassis (340-AKRZ) 1 Intel X710 Quad Port 10Gb DA/SFP+ Ethernet, Network Daughter Card (555-BCKP) 1 SAS 12Gbps HBA External Controller, Low Profile (405-AAFB) 1 iDRAC8 Enterprise, integrated Dell Remote Access Controller, Enterprise (385-BBHO) 1 Chassis with up to 8, 2.5" Hard Drives, 3 PCIe Slots (321-BBKL) 1 Brand/No-Bezel, OEM PowerEdge R630, 8 Drive Chassis (325-BBJC) 1 Performance BIOS Settings (384-BBBL) 1 RAID 1 for H330/H730/H730P (2 HDDs or SSDs) (780-BBJK) 1 PERC H330 Integrated RAID Controller (405-AAEF) 1 Intel Xeon E5-2680 v4 2.4GHz,35M Cache,9.60GT/s QPI,Turbo,HT,14C/28T (120W) Max Mem 2400MHz (338-BJDO) 1 Intel Xeon E5-2680 v4 2.4GHz,35M Cache,9.60GT/s QPI,Turbo,HT,14C/28T (120W) Max Mem 2400MHz (338-BJEE) 1 32GB RDIMM, 2400MT/s, Dual Rank, x4 Data Width (370-ACNS) 8 2400MT/s RDIMMs (370-ACPH) 1 Performance Optimized (370-AAIP) 1 200GB Solid State Drive SATA Mix Use MLC 6Gpbs 2.5in Hot-plug Drive, S3610 (400-AIFV) 2 Description Quantity OEM Dell Storage MD1420, 24 HDs 2.5", 2Us (210-ADBU) 1 Dell Hardware Limited Warranty Plus On Site Service Initial 15 Months (977-8949) 1 Dell Hardware Limited Warranty Plus On Site Service Extended Months(s) (977-8950) 1 ProSupport : Next Business Day Onsite Service After Problem Diagnosis, 15 Months (977-8952) 1 ProSupport : Next Business Day Onsite Service After Problem Diagnosis, 24 Months Extended (977-8953) 1 ProSupport : 7x24 Technical Support, 39 Months (977-8966) 1 Thank you choosing Dell ProSupport. For tech support, visit http://www.dell.com/support or call 1-800- 945-3355 (989-3439) 1 On-Site Installation Declined (900-9997) 1 Declined Remote Consulting Service (973-2426) 1 US Order (332-1286) 1 OEM SHIP,MD1420,DAO (340-AKRE) 1 No Raid, no Controller (405-AAFC) 1 Brand/No-Bezel, 2U, OEM Storage (325-BBIQ) 1 Enclosure Management Module, dual (403-BBGM) 1 Hard Drive Filler 2.5in, single blank (400-AEPR) 10 200GB Solid State Drive SAS Write Intensive MLC 12Gbps 2.5in Hot-plug Drive, PX04SH (400-AMEY) 14 Rack rail, 2Us, Static (770-BBJE) 1 OEM MD1420 Regulatory label, Mexico (389-BEVW) 1 Power supply, AC 600W, Redundant (450-AEBJ) 1 C13 to C14, PDU Style, 12 AMP, 6.5 Feet (2m) Power Cord, North America (492-BBDI) 1 C13 to C14, PDU Style, 12 AMP, 6.5 Feet (2m) Power Cord, North America (492-BBDI) 1 12Gb HD-Mini to HD-Mini SAS cable, 0.5m (470-ABDN) 4 Solutions Program Support (927-3179) 1 Description Quantity OEM PowerEdge R630 (210-ADCY) 1 PowerEdge R630 Motherboard MLK (329-BCZI) 1 ProSupport : Next Business Day Onsite Service After Problem Diagnosis, 39 Months (976-9165) 1 ProSupport : 7x24 Technical Support, 39 Months (976-9172) 1 Dell Hardware Limited Warranty Plus On Site Service (976-9191) 1 Thank you choosing Dell ProSupport. For tech support, visit http://www.dell.com/support or call 1-800- 945-3355 (989-3439) 1 US Order (332-1286) 1 OEM PowerEdge R630 Shipping-8 Drive Chassis (340-AKRZ) 1 Intel X710 Quad Port 10Gb DA/SFP+ Ethernet, Network Daughter Card (555-BCKP) 1 SAS 12Gbps HBA External Controller, Low Profile (405-AAFB) 1 iDRAC8 Enterprise, integrated Dell Remote Access Controller, Enterprise (385-BBHO) 1 Chassis with up to 8, 2.5" Hard Drives, 3 PCIe Slots (321-BBKL) 1 Brand/No-Bezel, OEM PowerEdge R630, 8 Drive Chassis (325-BBJC) 1 Performance BIOS Settings (384-BBBL) 1 RAID 1 for H330/H730/H730P (2 HDDs or SSDs) (780-BBJK) 1 PERC H330 Integrated RAID Controller (405-AAEF) 1 Intel Xeon E5-2680 v4 2.4GHz,35M Cache,9.60GT/s QPI,Turbo,HT,14C/28T (120W) Max Mem 2400MHz (338-BJDO) 1 Intel Xeon E5-2680 v4 2.4GHz,35M Cache,9.60GT/s QPI,Turbo,HT,14C/28T (120W) Max Mem 2400MHz (338-BJEE) 1 32GB RDIMM, 2400MT/s, Dual Rank, x4 Data Width (370-ACNS) 8 2400MT/s RDIMMs (370-ACPH) 1 Performance Optimized (370-AAIP) 1 200GB Solid State Drive SATA Mix Use MLC 6Gpbs 2.5in Hot-plug Drive, S3610 (400-AIFV) 2 PE Server FIPS TPM 1.2,CC (461-AADP) 1 No Systems Documentation, No OpenManage DVD Kit (631-AACK) 1 No Optical Drive Internal for 8 HD Chassis (429-AAQN) 1 ReadyRails Sliding Rails Without Cable Management Arm (770-BBBC) 1 Dual, Hot-plug, Redundant Power Supply (1+1), 495W (450-ADWQ) 1 C13 to C14, PDU Style, 12 AMP, 6.5 Feet (2m) Power Cord, North America (492-BBDI) 2 No Operating System (619-ABVR) 1 No Media Required (421-5736) 1 DIMM Blanks for System with 2 Processors (370-ABWE) 1 120W Heatsink for PowerEdge R630 (412-AAEE) 1 120W Heatsink for PowerEdge R630 (412-AAEE) 1 Solutions Program Support (927-3179) 1Description Quantity OEM Dell Storage MD1420, 24 HDs 2.5", 2Us (210-ADBU) 1 Dell Hardware Limited Warranty Plus On Site Service Initial 15 Months (977-8949) 1 Dell Hardware Limited Warranty Plus On Site Service Extended Months(s) (977-8950) 1 ProSupport : Next Business Day Onsite Service After Problem Diagnosis, 15 Months (977-8952) 1 ProSupport : Next Business Day Onsite Service After Problem Diagnosis, 24 Months Extended (977-8953) 1 ProSupport : 7x24 Technical Support, 39 Months (977-8966) 1 Thank you choosing Dell ProSupport. For tech support, visit http://www.dell.com/support or call 1-800- 945-3355 (989-3439) 1 On-Site Installation Declined (900-9997) 1 Declined Remote Consulting Service (973-2426) 1 US Order (332-1286) 1 OEM SHIP,MD1420,DAO (340-AKRE) 1 No Raid, no Controller (405-AAFC) 1 Brand/No-Bezel, 2U, OEM Storage (325-BBIQ) 1 Enclosure Management Module, dual (403-BBGM) 1 Hard Drive Filler 2.5in, single blank (400-AEPR) 10 200GB Solid State Drive SAS Write Intensive MLC 12Gbps 2.5in Hot-plug Drive, PX04SH (400-AMEY) 14 Rack rail, 2Us, Static (770-BBJE) 1 OEM MD1420 Regulatory label, Mexico (389-BEVW) 1 Power supply, AC 600W, Redundant (450-AEBJ) 1 C13 to C14, PDU Style, 12 AMP, 6.5 Feet (2m) Power Cord, North America (492-BBDI) 1 C13 to C14, PDU Style, 12 AMP, 6.5 Feet (2m) Power Cord, North America (492-BBDI) 1 12Gb HD-Mini to HD-Mini SAS cable, 0.5m (470-ABDN) 4 Solutions Program Support (927-3179) 1 Description Quantity OEM PowerEdge R630 (210-ADCY) 1 PowerEdge R630 Motherboard MLK (329-BCZI) 1 ProSupport : Next Business Day Onsite Service After Problem Diagnosis, 39 Months (976-9165) 1 ProSupport : 7x24 Technical Support, 39 Months (976-9172) 1 Dell Hardware Limited Warranty Plus On Site Service (976-9191) 1 Thank you choosing Dell ProSupport. For tech support, visit http://www.dell.com/support or call 1-800- 945-3355 (989-3439) 1 US Order (332-1286) 1 OEM PowerEdge R630 Shipping-8 Drive Chassis (340-AKRZ) 1 Intel X710 Quad Port 10Gb DA/SFP+ Ethernet, Network Daughter Card (555-BCKP) 1 SAS 12Gbps HBA External Controller, Low Profile (405-AAFB) 1 iDRAC8 Enterprise, integrated Dell Remote Access Controller, Enterprise (385-BBHO) 1 Chassis with up to 8, 2.5" Hard Drives, 3 PCIe Slots (321-BBKL) 1 Brand/No-Bezel, OEM PowerEdge R630, 8 Drive Chassis (325-BBJC) 1 Performance BIOS Settings (384-BBBL) 1 RAID 1 for H330/H730/H730P (2 HDDs or SSDs) (780-BBJK) 1 PERC H330 Integrated RAID Controller (405-AAEF) 1 Intel Xeon E5-2680 v4 2.4GHz,35M Cache,9.60GT/s QPI,Turbo,HT,14C/28T (120W) Max Mem 2400MHz (338-BJDO) 1 Intel Xeon E5-2680 v4 2.4GHz,35M Cache,9.60GT/s QPI,Turbo,HT,14C/28T (120W) Max Mem 2400MHz (338-BJEE) 1 32GB RDIMM, 2400MT/s, Dual Rank, x4 Data Width (370-ACNS) 8 2400MT/s RDIMMs (370-ACPH) 1 Performance Optimized (370-AAIP) 1 200GB Solid State Drive SATA Mix Use MLC 6Gpbs 2.5in Hot-plug Drive, S3610 (400-AIFV) 2 PE Server FIPS TPM 1.2,CC (461-AADP) 1 No Systems Documentation, No OpenManage DVD Kit (631-AACK) 1 No Optical Drive Internal for 8 HD Chassis (429-AAQN) 1 ReadyRails Sliding Rails Without Cable Management Arm (770-BBBC) 1 Dual, Hot-plug, Redundant Power Supply (1+1), 495W (450-ADWQ) 1 C13 to C14, PDU Style, 12 AMP, 6.5 Feet (2m) Power Cord, North America (492-BBDI) 2 No Operating System (619-ABVR) 1 No Media Required (421-5736) 1 DIMM Blanks for System with 2 Processors (370-ABWE) 1 120W Heatsink for PowerEdge R630 (412-AAEE) 1 120W Heatsink for PowerEdge R630 (412-AAEE) 1 Solutions Program Support (927-3179) 1
Created attachment 226501 [details] grep ACPI /var/log/messages output It appears that messages were rate limited in /var/log/messages. There were 17094100 objects created and 99% were in use. Over time the percent used drops as shown in previous comments
This output is very useful. It will help us to develop another patch that tells us exactly what control method(s) are executing. thanks.
Created attachment 226561 [details] Print ACPI Methods associated with Acpi-State. This patch prints the ACPI Method that's associated with the PSCOPE and WSCOPE descriptors that are being constantly generated as determined by the previous debug data. This should significantly narrow down the culprit. Again you'll want to clear dmesg just before running your script.
Any news on this?
We have had to wait for an internal re-organization to settle out before we can restart our investigations.
Ping...
I need to get resources assigned for reproducing this bug. Is the patch (above) the only thing you want changed?
Yes. This won't fix it. But it will hopefully show us the ASL method at the center of this problem.
I have applied the second patch and have reproduced the fault, but dmesg is silent during the run. Did you want both patches applied?
Refiling this bug to ACPICA bugzilla, please track it there. Thanks Lv