Re: Bisected regression in v5.1 on PowerBook G3 (Wallstreet)
On Wed, 26 Jun 2019, Christophe Leroy wrote: > Hi Finn, > > On 06/26/2019 02:06 AM, Finn Thain wrote: > > Hi Christophe, > > > > I received a report of a regression between v5.0 and v5.1 which causes > > the current release to crash during boot with a machine check > > exception. Please see console log below. > > > > Stan (whom I've Cc'd) tells me that this happens on every attempt to > > boot. I asked him to try 'git bisect'. The results are given below. > > Can you see anything in commit 93c4a162b014 that might explain this? > > Might be a false positive. That commit has a problem, but that problem > is fixed by 4622a2d43101 ("powerpc/6xx: fix setup and use of > SPRN_SPRG_PGDIR for hash32") > > I would bet your problem is related to commit f7354ccac844 ("powerpc/32: > Remove CURRENT_THREAD_INFO and rename TI_CPU"). That problem is fixed by > commit 397d2300b08c ("powerpc/32s: fix flush_hash_pages() on SMP") > upstream, and in linux 5.1.4 by commit fda49aec2515 on > stable/linux-5.1.y > I see. I've just discovered that this issue has already been covered on this list. I should have done a bit more research. > Can you test ? > Stan did some more tests and confirmed that the problem has been fixed in 397d2300b08c and stable/linux-5.1.y. Thanks. -- > Thanks > Christophe >
Bisected regression in v5.1 on PowerBook G3 (Wallstreet)
Hi Christophe, I received a report of a regression between v5.0 and v5.1 which causes the current release to crash during boot with a machine check exception. Please see console log below. Stan (whom I've Cc'd) tells me that this happens on every attempt to boot. I asked him to try 'git bisect'. The results are given below. Can you see anything in commit 93c4a162b014 that might explain this? I can also provide the .config if it would help. $ cat bisect.log 5.0.0-pmac-ide-03515-g3478588b5136 #2 worked 5.0.0-pmac-ide-05504-gda2577fe63f8 #3 worked 5.0.0-pmac-ide-06224-g67e79a6dc266 #4 worked 5.0.0-pmac-ide-06622-g1fc1cd8399ab #5 worked 5.0.0-rc2-pmac-ide-00215=g9580b71b5a78 #6 failed 5.0.0-rc2-pmac-ide-00113-gfe1ef6bcdb4f #7 worked 5.0.0-rc2-pmac-ide-00164-gd5f17ee96447 #8 failed 5.0.0-rc2-pmac-ide-00138-g84de6ab0e904 #9 failed 5.0.0-rc2-pmac-ide-00125-ge995265252fa #10 worked 5.0.0-rc2-pmac-ide-00131-g93c4a162b014 #11 failed 5.0.0-rc2-pmac-ide-00128-g36da5ff0bea2 #12 worked 93c4a162b014d238a287f8264adb25c009c79e61 is the first bad commit commit 93c4a162b014d238a287f8264adb25c009c79e61 Author: Christophe Leroy Date: Thu Feb 21 10:37:55 2019 + powerpc/6xx: Store PGDIR physical address in a SPRG Use SPRN_SPRG2 to store the current thread PGDIR and avoid reading thread_struct.pgdir at every TLB miss. Signed-off-by: Christophe Leroy Signed-off-by: Michael Ellerman :04 04 dcd7171dff5ba5bf895e4399d9d859c91c5a8293 c51e7def7720499289420ace421cf755bf3bf37e M arch [0.00] printk: debug: ignoring loglevel setting. [0.00] Total memory = 512MB; using 1024kB for hash table (at (ptrval)) [0.00] Linux version 5.1.0-pmac-ide (fthain@nippy) (gcc version 4.6.4 (btc)) #1 SMP Sun Jun 23 14:46:26 AEST 2019 [0.00] Found a Gatwick mac-io controller, rev: 0, mapped at 0x(ptrval) [0.00] Found a Heathrow mac-io controller, rev: 0, mapped at 0x(ptrval) [0.00] PowerMac motherboard: PowerBook Wallstreet [0.00] PMU driver v2 initialized for PowerBook G3 Series, firmware: 0a [0.00] Using PowerMac machine description [0.00] printk: bootconsole [udbg0] enabled [0.00] CPU maps initialized for 1 thread per core [0.00] (thread shift is 0) [0.00] - [0.00] Hash_size = 0x10 [0.00] phys_mem_size = 0x2000 [0.00] dcache_bsize = 0x20 [0.00] icache_bsize = 0x20 [0.00] cpu_features = 0x0501a008 [0.00] possible= 0x2f7ff14b [0.00] always = 0x0100 [0.00] cpu_user_features = 0x8c01 0x [0.00] mmu_features = 0x0001 [0.00] Hash = 0x(ptrval) [0.00] Hash_mask = 0x3fff [0.00] - [0.00] Found Grackle (MPC106) PCI host bridge at 0x8000. Firmware bus number: 0->0 [0.00] PCI host bridge /pci (primary) ranges: [0.00] IO 0xfe00..0xfe7f -> 0x [0.00] MEM 0xfd00..0xfdff -> 0x [0.00] MEM 0x8000..0xfcff -> 0x8000 [0.00] nvram: OF partition at 0x1800 [0.00] nvram: XP partition at 0x1300 [0.00] nvram: NR partition at 0x1400 [0.00] Top of RAM: 0x2000, Total RAM: 0x2000 [0.00] Memory hole size: 0MB [0.00] Zone ranges: [0.00] Normal [mem 0x-0x1fff] [0.00] HighMem empty [0.00] Movable zone start for each node [0.00] Early memory node ranges [0.00] node 0: [mem 0x-0x1fff] [0.00] Initmem setup node 0 [mem 0x-0x1fff] [0.00] On node 0 totalpages: 131072 [0.00] Normal zone: 1024 pages used for memmap [0.00] Normal zone: 0 pages reserved [0.00] Normal zone: 131072 pages, LIFO batch:31 [0.00] percpu: Embedded 14 pages/cpu s24972 r8192 d24180 u57344 [0.00] pcpu-alloc: s24972 r8192 d24180 u57344 alloc=14*4096 [0.00] pcpu-alloc: [0] 0 [0] 1 [0.00] Built 1 zonelists, mobility grouping on. Total pages: 130048 [0.00] Kernel command line: root=/dev/hda11 video=atyfb:vmode:14,cmode:32 ignore_loglevel printk.time console=ttyS0,9600n8 console=tty [0.00] Dentry cache hash table entries: 65536 (order: 6, 262144 bytes) [0.00] Inode-cache hash table entries: 32768 (order: 5, 131072 bytes) [0.00] Memory: 503332K/524288K available (5468K kernel code, 272K rwdata, 1364K rodata, 264K init, 182K bss, 20956K reserved, 0K cma-reserved, 0K highmem) [0.00] Kernel virtual memory layout: [0.00] * 0xfffbf000..0xf000 : fixmap [0.00] *
Re: Bisected regression in v5.1 on PowerBook G3 (Wallstreet)
Hi Finn, On 06/26/2019 02:06 AM, Finn Thain wrote: Hi Christophe, I received a report of a regression between v5.0 and v5.1 which causes the current release to crash during boot with a machine check exception. Please see console log below. Stan (whom I've Cc'd) tells me that this happens on every attempt to boot. I asked him to try 'git bisect'. The results are given below. Can you see anything in commit 93c4a162b014 that might explain this? Might be a false positive. That commit has a problem, but that problem is fixed by 4622a2d43101 ("powerpc/6xx: fix setup and use of SPRN_SPRG_PGDIR for hash32") I would bet your problem is related to commit f7354ccac844 ("powerpc/32: Remove CURRENT_THREAD_INFO and rename TI_CPU"). That problem is fixed by commit 397d2300b08c ("powerpc/32s: fix flush_hash_pages() on SMP") upstream, and in linux 5.1.4 by commit fda49aec2515 on stable/linux-5.1.y Can you test ? Thanks Christophe I can also provide the .config if it would help. $ cat bisect.log 5.0.0-pmac-ide-03515-g3478588b5136 #2 worked 5.0.0-pmac-ide-05504-gda2577fe63f8 #3 worked 5.0.0-pmac-ide-06224-g67e79a6dc266 #4 worked 5.0.0-pmac-ide-06622-g1fc1cd8399ab #5 worked 5.0.0-rc2-pmac-ide-00215=g9580b71b5a78 #6 failed 5.0.0-rc2-pmac-ide-00113-gfe1ef6bcdb4f #7 worked 5.0.0-rc2-pmac-ide-00164-gd5f17ee96447 #8 failed 5.0.0-rc2-pmac-ide-00138-g84de6ab0e904 #9 failed 5.0.0-rc2-pmac-ide-00125-ge995265252fa #10 worked 5.0.0-rc2-pmac-ide-00131-g93c4a162b014 #11 failed 5.0.0-rc2-pmac-ide-00128-g36da5ff0bea2 #12 worked 93c4a162b014d238a287f8264adb25c009c79e61 is the first bad commit commit 93c4a162b014d238a287f8264adb25c009c79e61 Author: Christophe Leroy Date: Thu Feb 21 10:37:55 2019 + powerpc/6xx: Store PGDIR physical address in a SPRG Use SPRN_SPRG2 to store the current thread PGDIR and avoid reading thread_struct.pgdir at every TLB miss. Signed-off-by: Christophe Leroy Signed-off-by: Michael Ellerman :04 04 dcd7171dff5ba5bf895e4399d9d859c91c5a8293 c51e7def7720499289420ace421cf755bf3bf37e M arch [0.00] printk: debug: ignoring loglevel setting. [0.00] Total memory = 512MB; using 1024kB for hash table (at (ptrval)) [0.00] Linux version 5.1.0-pmac-ide (fthain@nippy) (gcc version 4.6.4 (btc)) #1 SMP Sun Jun 23 14:46:26 AEST 2019 [0.00] Found a Gatwick mac-io controller, rev: 0, mapped at 0x(ptrval) [0.00] Found a Heathrow mac-io controller, rev: 0, mapped at 0x(ptrval) [0.00] PowerMac motherboard: PowerBook Wallstreet [0.00] PMU driver v2 initialized for PowerBook G3 Series, firmware: 0a [0.00] Using PowerMac machine description [0.00] printk: bootconsole [udbg0] enabled [0.00] CPU maps initialized for 1 thread per core [0.00] (thread shift is 0) [0.00] - [0.00] Hash_size = 0x10 [0.00] phys_mem_size = 0x2000 [0.00] dcache_bsize = 0x20 [0.00] icache_bsize = 0x20 [0.00] cpu_features = 0x0501a008 [0.00] possible= 0x2f7ff14b [0.00] always = 0x0100 [0.00] cpu_user_features = 0x8c01 0x [0.00] mmu_features = 0x0001 [0.00] Hash = 0x(ptrval) [0.00] Hash_mask = 0x3fff [0.00] - [0.00] Found Grackle (MPC106) PCI host bridge at 0x8000. Firmware bus number: 0->0 [0.00] PCI host bridge /pci (primary) ranges: [0.00] IO 0xfe00..0xfe7f -> 0x [0.00] MEM 0xfd00..0xfdff -> 0x [0.00] MEM 0x8000..0xfcff -> 0x8000 [0.00] nvram: OF partition at 0x1800 [0.00] nvram: XP partition at 0x1300 [0.00] nvram: NR partition at 0x1400 [0.00] Top of RAM: 0x2000, Total RAM: 0x2000 [0.00] Memory hole size: 0MB [0.00] Zone ranges: [0.00] Normal [mem 0x-0x1fff] [0.00] HighMem empty [0.00] Movable zone start for each node [0.00] Early memory node ranges [0.00] node 0: [mem 0x-0x1fff] [0.00] Initmem setup node 0 [mem 0x-0x1fff] [0.00] On node 0 totalpages: 131072 [0.00] Normal zone: 1024 pages used for memmap [0.00] Normal zone: 0 pages reserved [0.00] Normal zone: 131072 pages, LIFO batch:31 [0.00] percpu: Embedded 14 pages/cpu s24972 r8192 d24180 u57344 [0.00] pcpu-alloc: s24972 r8192 d24180 u57344 alloc=14*4096 [0.00] pcpu-alloc: [0] 0 [0] 1 [0.00] Built 1 zonelists, mobility grouping on. Total pages: 130048 [0.00] Kernel command line: