Date: Tue, 22 Sep 2020 12:35:26 -0700 From: Mark Millard <marklmi@yahoo.com> To: Brandon Bergren <bdragon@FreeBSD.org> Cc: FreeBSD PowerPC ML <freebsd-ppc@freebsd.org> Subject: Re: head -r365932 on PowerMac G5 (2 dual-core sockets): Crashes before login prompt if powerd is enabled in /etc/rc.conf Message-ID: <B4D16405-1B0C-4500-866F-75F32FED235A@yahoo.com> In-Reply-To: <E23CF7F6-B69C-44C7-B356-4AB65E011268@yahoo.com> References: <52783D16-5DCA-45BC-9238-2518326454A1@yahoo.com> <6E99EE39-D2B8-415A-A5BF-823C0F0C22D6@yahoo.com> <cd9d2b72-219f-4550-a437-4ac3aa1da66d@www.fastmail.com> <AF27169A-00FC-4984-83C2-307EA885D7A1@yahoo.com> <11fe573a-24c3-47be-95ed-c601ec54f168@www.fastmail.com> <E23CF7F6-B69C-44C7-B356-4AB65E011268@yahoo.com>
index | next in thread | previous in thread | raw e-mail
On 2020-Sep-22, at 12:15, Mark Millard <marklmi at yahoo.com> wrote: > On 2020-Sep-22, at 12:00, Brandon Bergren <bdragon at FreeBSD.org> wrote: > >> Weird, that read_scom has the "inlines not being inlined" problem. >> >> Are you sure your local tree has my https://svnweb.freebsd.org/base/head/?view=revision&revision=365441 fixes? > > # svnlite diff /usr/src/sys/conf/kern.pre.mk > # svnlite info /usr/src > Path: . > Working Copy Root Path: /usr/src > URL: https://svn.freebsd.org/base/head > Relative URL: ^/head > Repository Root: https://svn.freebsd.org/base > Repository UUID: ccf9f872-aa2e-dd11-9fc8-001c23d0bc1f > Revision: 365932 > Node Kind: directory > Schedule: normal > Last Changed Author: jrtc27 > Last Changed Rev: 365932 > Last Changed Date: 2020-09-20 16:20:18 -0700 (Sun, 20 Sep 2020) > > But it used -O instead of -O2 . . . > > # Meta data file /usr/obj/powerpc64vtsc_clang/powerpc.powerpc64/usr/src/powerpc.powerpc64/sys/GENERIC64vtsc-NODBG/pcr.o.meta > CMD cc -target powerpc64-unknown-freebsd13.0 --sysroot=/usr/obj/powerpc64vtsc_clang/powerpc.powerpc64/usr/src/powerpc.powerpc64/tmp -B/usr/obj/powerpc64vtsc_clang/powerpc.powerpc64/usr/src/powerpc.powerpc64/tmp/usr/bin -c -O -pipe -g -nostdinc -I. -I/usr/src/sys -I/usr/src/sys/contrib/ck/include -I/usr/src/sys/contrib/libfdt -D_KERNEL -DHAVE_KERNEL_OPTION_HEADERS -include opt_global.h -fno-common -msoft-float -fPIC -fno-omit-frame-pointer -fdebug-prefix-map=./machine=/usr/src/sys/powerpc/include -mno-altivec -msoft-float -mabi=elfv2 -ffreestanding -fwrapv -fstack-protector -gdwarf-2 -Wall -Wredundant-decls -Wnested-externs -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith -Wcast-qual -Wundef -Wno-pointer-sign -D__printf__=__freebsd_kprintf__ -Wmissing-include-dirs -fdiagnostics-show-option -Wno-unknown-pragmas -Wno-error-tautological-compare -Wno-error-empty-body -Wno-error-parentheses-equality -Wno-error-unused-function -Wno-error-pointer-sign -Wno-error-shift-negative-value -Wno-address-of-packed-member -Wno-format-zero-length -std=iso9899:1999 /usr/src/sys/powerpc/cpufreq/pcr.c > CMD ctfconvert -L VERSION -g pcr.o > CWD /usr/obj/powerpc64vtsc_clang/powerpc.powerpc64/usr/src/powerpc.powerpc64/sys/GENERIC64vtsc-NODBG > TARGET pcr.o > OODATE offset.inc assym.inc The logic: .if defined(DEBUG) .if ${MACHINE_ARCH} == "powerpc" || ${MACHINE_ARCH} == "powerpcspe" # Work around clang 11 miscompile on 32 bit powerpc. _MINUS_O= -O2 .else _MINUS_O= -O .endif CTFFLAGS+= -g .else _MINUS_O= -O2 .endif Seems to target 32 bit powerpc but the context I'm reporting is for powerpc64 overall. May be the mis-compiles are not limited to 32-bit powerpc variants? In fact I have DEBUG defined despite not having witness and the like. I include GENERIC64 and the am explicit about overriding things that I care about, even some things that are the same as in GENERIC64. # more /usr/src/sys/powerpc/conf/GENERIC64vtsc-NODBG # # GENERIC -- Custom configuration for the powerpc/powerpc64 # include "GENERIC64" ident GENERIC64vtsc-NODBG makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols nooptions PS3 # Sony Playstation 3 HACK!!! to allow sc options KDB # Enable kernel debugger support # For minimum debugger support (stable branch) use: options KDB_TRACE # Print a stack trace for a panic options DDB # Enable the kernel debugger options GDB # HACK!!! ... options ALT_BREAK_TO_DEBUGGER options BREAK_TO_DEBUGGER # Extra stuff: #options VERBOSE_SYSINIT=0 # Enable verbose sysinit messages #options BOOTVERBOSE=1 #options BOOTHOWTO=RB_VERBOSE #options KTR #options KTR_MASK=KTR_TRAP ##options KTR_CPUMASK=0xF #options KTR_VERBOSE # HACK!!! to allow sc for 2560x1440 display on Radeon X1950 that vt historically mishandled during booting device sc #device kbdmux # HACK: already listed by vt options SC_OFWFB # OFW frame buffer options SC_DFLT_FONT # compile font in makeoptions SC_DFLT_FONT=cp437 # Disable any extra checking for. . . nooptions DEADLKRES # Enable the deadlock resolver nooptions INVARIANTS # Enable calls of extra sanity checking nooptions INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS nooptions WITNESS # Enable checks to detect deadlocks and cycles nooptions WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed nooptions DIAGNOSTIC nooptions MALLOC_DEBUG_MAXZONES # Separate malloc(9) zones # Avoid dynamic loads? device filemon device geom_label device mac_ntpd # Avoid .ko dymaical loads that may mix dpcpu_off[cpuid] use with # internal pcpu_entry_NAME things: dpcpu_off[cpuid] use attempts # to convert 0xc???_????_????_???? to some 0xe???_????_????_???? # by addition but the dynamically loaded pcpu_entry_NAME is # already at some 0xe???_????_????_???? and the addition # overflows/truncates, producing rejected 0x0???_????_????_???? # addresses. #device epair #options IPFIREWALL # ipfw.ko I've no interest in this #options compat_linux # no longer possible to statically include? (32 and 64 bit mixed support?) #device linuxkpi # linuxkpi.ko I've no interest in #device siftr # siftr.ko: TCP statistics that I've no interest in >> On Tue, Sep 22, 2020, at 1:55 PM, Mark Millard wrote: >>> >>> >>> On 2020-Sep-22, at 08:58, Brandon Bergren <bdragon@FreeBSD.org> wrote: >>> >>>> In theory, this would also crash if you did "sysctl dev.cpu.0.freq". >>> >>> Yep, that dies with a backtrace but not taking input to >>> the db> prompt. >>> >>> The call stack looks to have the same sequence. >>> >>> Different srr0 and lr values were listed when I tried >>> that: >>> >>> srr0=0x8004'4000'0000'0000 (0xc004'4000'0000'0000) >>> . . . >>> lr =0x8004'4000'0000'0000 (0xc004'4000'0000'0000) >>> >>> >>>> You sure the lack of a backtrace isn't just that you are using a nodebug config? >>> >>> Back when I was using the FireWire based debugging I >>> discovered that it continued to report material after >>> the monitor updates stopped. I learned to not use >>> the last message on screen to guess where it was >>> having a problem, other than "sometime after the >>> last message shown". >>> >>>> Could you please disassemble read_scom? >>> >>> Sure: >>> >>> 0000000000ad7f24 <read_scom> addis r2,r12,132 >>> 0000000000ad7f28 <read_scom+0x4> addi r2,r2,220 >>> 0000000000ad7f2c <read_scom+0x8> mflr r0 >>> 0000000000ad7f30 <read_scom+0xc> std r31,-8(r1) >>> 0000000000ad7f34 <read_scom+0x10> std r0,16(r1) >>> 0000000000ad7f38 <read_scom+0x14> stdu r1,-64(r1) >>> 0000000000ad7f3c <read_scom+0x18> mr r31,r1 >>> 0000000000ad7f40 <read_scom+0x1c> std r29,40(r31) >>> 0000000000ad7f44 <read_scom+0x20> std r30,48(r31) >>> 0000000000ad7f48 <read_scom+0x24> bl 0000000000ad7e58 <mfmsr> >>> 0000000000ad7f4c <read_scom+0x28> mr r30,r3 >>> 0000000000ad7f50 <read_scom+0x2c> rldicl r3,r3,48,1 >>> 0000000000ad7f54 <read_scom+0x30> rotldi r3,r3,16 >>> 0000000000ad7f58 <read_scom+0x34> bl 0000000000ad7e6c <mtmsr> >>> 0000000000ad7f5c <read_scom+0x38> bl 0000000000ad7e84 <isync> >>> 0000000000ad7f60 <read_scom+0x3c> lis r3,16512 >>> 0000000000ad7f64 <read_scom+0x40> ori r3,r3,33024 >>> 0000000000ad7f68 <read_scom+0x44> mtspr 276,r3 >>> 0000000000ad7f6c <read_scom+0x48> bl 0000000000ad7e84 <isync> >>> 0000000000ad7f70 <read_scom+0x4c> mfspr r29,277 >>> 0000000000ad7f74 <read_scom+0x50> mr r30,r29 >>> 0000000000ad7f78 <read_scom+0x54> rldicl r29,r29,32,32 >>> 0000000000ad7f7c <read_scom+0x58> mfspr r3,276 >>> 0000000000ad7f80 <read_scom+0x5c> mr r3,r30 >>> 0000000000ad7f84 <read_scom+0x60> bl 0000000000ad7e6c <mtmsr> >>> 0000000000ad7f88 <read_scom+0x64> bl 0000000000ad7e84 <isync> >>> 0000000000ad7f8c <read_scom+0x68> mr r3,r29 >>> 0000000000ad7f90 <read_scom+0x6c> ld r30,48(r31) >>> 0000000000ad7f94 <read_scom+0x70> ld r29,40(r31) >>> 0000000000ad7f98 <read_scom+0x74> addi r1,r1,64 >>> 0000000000ad7f9c <read_scom+0x78> ld r0,16(r1) >>> 0000000000ad7fa0 <read_scom+0x7c> mtlr r0 >>> 0000000000ad7fa4 <read_scom+0x80> ld r31,-8(r1) >>> 0000000000ad7fa8 <read_scom+0x84> blr >>> >>> >>>> On Tue, Sep 22, 2020, at 12:46 AM, Mark Millard via freebsd-ppc wrote: >>>>> >>>>> >>>>> On 2020-Sep-21, at 21:34, Mark Millard <marklmi at yahoo.com> wrote: >>>>> >>>>>> This was discovered while doing a head -r363590 -> -r365932 >>>>>> upgrade to FreeBSD. (A non-debug system context.) >>>>>> >>>>>> It first showed up only having updated the kernel. It still >>>>>> shows up after updating world as well. It is now running: >>>>>> >>>>>> # uname -apKU >>>>>> FreeBSD FBSDG5L2 13.0-CURRENT FreeBSD 13.0-CURRENT #16 r365932M: Sun Sep 20 19:57:07 PDT 2020 root@FBSDFHUGE:/usr/obj/powerpc64vtsc_clang/powerpc.powerpc64/usr/src/powerpc.powerpc64/sys/GENERIC64vtsc-NODBG powerpc powerpc64 1300115 1300115 >>>>>> >>>>>> but with /etc/rc.conf having powerd disabled: >>>>>> >>>>>> #powerd_enable="YES" >>>>>> >>>>>> The crash now is now silent, not getting to the db> prompt >>>>>> and not showing any messages or backtrace. >>>>>> >>>>>> Prior to world being updated it crashed with a traceback. >>>>>> A quick summary from a camera picture: >>>>>> >>>>>> fatal kernel trap: >>>>>> . . . >>>>>> pid = 1126, comm = powerd >>>>>> . . . >>>>>> kernel PGM trap by 0: . . . >>>>>> at pcr_get+0x4c >>>>>> at CPUFREQ_DRV_GET+0x78 >>>>>> at cpufreq_get_frequency+0x20 >>>>>> at cpufreq_get_level+0x2c >>>>>> at cf_get_method+0x20c >>>>>> at CPUFREQ_GET+0x78 >>>>>> at cpufreq_curr_sysctl+0x70 >>>>>> at sysctl_root_handler_locked+0x10c >>>>>> at sysctl_root+0x26c >>>>>> at userland_sysctl+0x14c >>>>>> at sys___sysctl+0x8c >>>>>> at syscallenter+0x188 >>>>>> at syscall+0x60 >>>>>> at trap+0x498 >>>>>> at powerpc_interrrupt+0x110 >>>>>> user SC trap . . . >>>>>> >>>>>> After this I tried to make a dump and then proceeded >>>>>> with disabling powerd in /etc/rc.conf and doing the >>>>>> world update. >>>>>> >>>>>> Unfortunately, while a dump was written, the core.txt >>>>>> file from the -r365932 world boot that processed the >>>>>> dump reported "invalid corefile" all over the place. >>>>>> >>>>>> With powerpd disabled the G5 seems to be operational. >>>>>> But turning powerd back on in /etc/rc.conf and rebooting >>>>>> prevents the boot from completing, no messages, no >>>>>> db> prompt. So I now leave powerd disabled. >>>>> >>>>> Some additional low-level information: >>>>> >>>>> For exception 0x700 (program) the screen picture >>>>> shows (but I've added ' use): >>>>> >>>>> srr0=0x0 (0x4000'0000'0000'0000) >>>>> . . . >>>>> lr =0x0 (0x4000'0000'0000'0000) >>>>> >>>>> The kernel PGM trap notice does report: >>>>> >>>>> ctr=0xc000'0000'00ad'7ad4 >>>>> (the start of pcr_get but with the 0xc >>>>> prefix) >>>>> >>>>> I'll remind of the pcr_get+0x4c report in the summary. >>>>> >>>>> objdump for /boot/kernel/kernel reports: >>>>> >>>>> 0000000000ad7ad4 <pcr_get> addis r2,r12,132 >>>>> 0000000000ad7ad8 <pcr_get+0x4> addi r2,r2,1324 >>>>> 0000000000ad7adc <pcr_get+0x8> cmpldi r4,0 >>>>> 0000000000ad7ae0 <pcr_get+0xc> beq 0000000000ad7b48 <pcr_get+0x74> >>>>> 0000000000ad7ae4 <pcr_get+0x10> mflr r0 >>>>> 0000000000ad7ae8 <pcr_get+0x14> std r31,-8(r1) >>>>> 0000000000ad7aec <pcr_get+0x18> std r0,16(r1) >>>>> 0000000000ad7af0 <pcr_get+0x1c> stdu r1,-64(r1) >>>>> 0000000000ad7af4 <pcr_get+0x20> mr r31,r1 >>>>> 0000000000ad7af8 <pcr_get+0x24> std r29,40(r31) >>>>> 0000000000ad7afc <pcr_get+0x28> mr r29,r3 >>>>> 0000000000ad7b00 <pcr_get+0x2c> li r3,-1 >>>>> 0000000000ad7b04 <pcr_get+0x30> std r30,48(r31) >>>>> 0000000000ad7b08 <pcr_get+0x34> mr r30,r4 >>>>> 0000000000ad7b0c <pcr_get+0x38> std r3,32(r4) >>>>> 0000000000ad7b10 <pcr_get+0x3c> std r3,24(r4) >>>>> 0000000000ad7b14 <pcr_get+0x40> std r3,16(r4) >>>>> 0000000000ad7b18 <pcr_get+0x44> std r3,8(r4) >>>>> 0000000000ad7b1c <pcr_get+0x48> std r3,0(r4) >>>>> 0000000000ad7b20 <pcr_get+0x4c> bl 0000000000ad7f2c <read_scom+0x8> >>>>> 0000000000ad7b24 <pcr_get+0x50> rldicl r3,r3,8,62 >>>>> 0000000000ad7b28 <pcr_get+0x54> li r4,10000 >>>>> 0000000000ad7b2c <pcr_get+0x58> stw r4,0(r30) >>>>> 0000000000ad7b30 <pcr_get+0x5c> cmpldi r3,1 >>>>> 0000000000ad7b34 <pcr_get+0x60> beq 0000000000ad7b50 <pcr_get+0x7c> >>>>> 0000000000ad7b38 <pcr_get+0x64> cmpldi r3,2 >>>>> 0000000000ad7b3c <pcr_get+0x68> bne 0000000000ad7b58 <pcr_get+0x84> >>>>> 0000000000ad7b40 <pcr_get+0x6c> li r3,2500 >>>>> 0000000000ad7b44 <pcr_get+0x70> b 0000000000ad7b54 <pcr_get+0x80> >>>>> 0000000000ad7b48 <pcr_get+0x74> li r3,22 >>>>> 0000000000ad7b4c <pcr_get+0x78> blr >>>>> 0000000000ad7b50 <pcr_get+0x7c> li r3,5000 >>>>> 0000000000ad7b54 <pcr_get+0x80> stw r3,0(r30) >>>>> 0000000000ad7b58 <pcr_get+0x84> std r29,16(r30) >>>>> 0000000000ad7b5c <pcr_get+0x88> ld r30,48(r31) >>>>> 0000000000ad7b60 <pcr_get+0x8c> ld r29,40(r31) >>>>> 0000000000ad7b64 <pcr_get+0x90> addi r1,r1,64 >>>>> 0000000000ad7b68 <pcr_get+0x94> ld r0,16(r1) >>>>> 0000000000ad7b6c <pcr_get+0x98> li r3,0 >>>>> 0000000000ad7b70 <pcr_get+0x9c> mtlr r0 >>>>> 0000000000ad7b74 <pcr_get+0xa0> ld r31,-8(r1) >>>>> 0000000000ad7b78 <pcr_get+0xa4> blr >> -- Brandon Bergren bdragon@FreeBSD.org === Mark Millard marklmi at yahoo.com ( dsl-only.net went away in early 2018-Mar) === Mark Millard marklmi at yahoo.com ( dsl-only.net went away in early 2018-Mar)home | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?B4D16405-1B0C-4500-866F-75F32FED235A>