Skip to contentSkip to conversation

Add command to display or save Linux PStore dumps

PATCH Superseded Frédéric Danis 6 participants 16 messages Mar 19, 2020 – Mar 23, 2020

v3 of 5 Superseded Latest is v5 →

lore
Share & tools
lore
#0 lore raw

v3 cover Add command to display or save Linux PStore dumps

This serie of patches adds a new pstore command allowing to display or save ramoops logs (oops, panic, console, ftrace and user) generated by a previous kernel crash.
PStore parameters can be set in U-Boot configuration file, or at run-time using "pstore set" command. For kernel using Device Tree, the parameters are dynamically added to Device Tree.
Records size should be the same as the ones used by kernel, and should be a power of 2.

Since v2:
- Add default value for PStore memory size
- Remove default value of PStore memory address
- Update config entry helps
- Replace calls to debug() by log_debug()
- Update documentation
- Replace 1M test file by 3 * 4K files and build pstore memory during test
- Add fdt_fixup_pstore() to pass PStore/Ramoops parameters to kernel

Since v1:
- Fix 64bit mode build warnings
- Add documentation
- Add function description comments
- Replace calls to pr_debug() by debug()
- Add CONFIG_CMD_PSTORE to sandbox and sandbox64
- Add unit tests

Fr?d?ric Danis (3):
cmd: Add command to display or save Linux PStore dumps
test: Add PStore command tests
cmd: Fixup DT to pass PStore Ramoops parameters

 cmd/Kconfig                                |  71 +++
 cmd/Makefile                               |   1 +
 cmd/pstore.c                               | 543 +++++++++++++++++++++
 common/image-fdt.c                         |   4 +
 configs/sandbox64_defconfig                |   1 +
 configs/sandbox_defconfig                  |   1 +
 doc/index.rst                              |   7 +
 doc/pstore.rst                             |  76 +++
 include/fdt_support.h                      |   3 +
 test/py/tests/test_pstore.py               |  73 +++
 test/py/tests/test_pstore_data_console.hex | Bin 0 -> 4096 bytes
 test/py/tests/test_pstore_data_panic1.hex  | Bin 0 -> 4096 bytes
 test/py/tests/test_pstore_data_panic2.hex  | Bin 0 -> 4096 bytes
 13 files changed, 780 insertions(+)
 create mode 100644 cmd/pstore.c
 create mode 100644 doc/pstore.rst
 create mode 100644 test/py/tests/test_pstore.py
 create mode 100644 test/py/tests/test_pstore_data_console.hex
 create mode 100644 test/py/tests/test_pstore_data_panic1.hex
 create mode 100644 test/py/tests/test_pstore_data_panic2.hex
Signature · 1 line
2.18.0
#1 lore raw

PATCH v3 1/3 cmd: Add command to display or save Linux PStore dumps

This patch adds a new pstore command allowing to display or save ramoops logs (oops, panic, console, ftrace and user) generated by a previous kernel crash.
PStore parameters can be set in U-Boot configuration file, or at run-time using "pstore set" command. Records size should be the same as the ones used by kernel, and should be a power of 2.
This command allows:
- to display uncompressed logs
- to save compressed or uncompressed logs, compressed logs are saved as a compressed stream, it may need some work to be able to decompress it, e.g. adding a fake header:
"printf "\x1f\x8b\x08\x00\x00\x00\x00\x00\x00\x00" |
cat - dmesg-ramoops-0.enc.z | gzip -dc"
- ECC part is not used to check memory corruption
- only 1st FTrace log is displayed or saved

---
 cmd/Kconfig    |  71 +++++++
 cmd/Makefile   |   1 +
 cmd/pstore.c   | 505 +++++++++++++++++++++++++++++++++++++++++++++++++
 doc/index.rst  |   7 +
 doc/pstore.rst |  74 ++++++++
 5 files changed, 658 insertions(+)
 create mode 100644 cmd/pstore.c
 create mode 100644 doc/pstore.rst
cmd/Kconfig +71 -0
index 6403bc45a5..cd202bf7fb 100644
@@ -1736,6 +1736,77 @@ config CMD_QFW
 	  feature is to allow easy loading of files passed to qemu-system
 	  via -kernel / -initrd
 
+config CMD_PSTORE
+	bool "pstore"
+	help
+	  This provides access to Linux PStore with Rammoops backend. The main
+	  feature is to allow to display or save PStore records.
+
+	  See doc/pstore.rst for more information.
+
+if CMD_PSTORE
+
+config CMD_PSTORE_MEM_ADDR
+	hex "Memory Address"
+	depends on CMD_PSTORE
+	help
+	  Base addr used for PStore ramoops memory, should be identical to
+	  ramoops.mem_address parameter used by kernel
+
+config CMD_PSTORE_MEM_SIZE
+	hex "Memory size"
+	depends on CMD_PSTORE
+	default "0x10000"
+	help
+	  Size of PStore ramoops memory, should be identical to ramoops.mem_size
+	  parameter used by kernel, a power of 2 and larger than the sum of the
+	  record sizes
+
+config CMD_PSTORE_RECORD_SIZE
+	hex "Dump record size"
+	depends on CMD_PSTORE
+	default "0x1000"
+	help
+	  Size of each dump done on oops/panic, should be identical to
+	  ramoops.record_size parameter used by kernel and a power of 2
+	  Must be non-zero
+
+config CMD_PSTORE_CONSOLE_SIZE
+	hex "Kernel console log size"
+	depends on CMD_PSTORE
+	default "0x1000"
+	help
+	  Size of kernel console log, should be identical to
+	  ramoops.console_size parameter used by kernel and a power of 2
+	  Must be non-zero
+
+config CMD_PSTORE_FTRACE_SIZE
+	hex "FTrace log size"
+	depends on CMD_PSTORE
+	default "0x1000"
+	help
+	  Size of ftrace log, should be identical to ramoops.ftrace_size
+	  parameter used by kernel and a power of 2
+
+config CMD_PSTORE_PMSG_SIZE
+	hex "User space message log size"
+	depends on CMD_PSTORE
+	default "0x1000"
+	help
+	  Size of user space message log, should be identical to
+	  ramoops.pmsg_size parameter used by kernel and a power of 2
+
+config CMD_PSTORE_ECC_SIZE
+	int "ECC size"
+	depends on CMD_PSTORE
+	default "0"
+	help
+	if non-zero, the option enables ECC support and specifies ECC buffer
+	size in bytes (1 is a special value, means 16 bytes ECC), should be
+	identical to ramoops.ramoops_ecc parameter used by kernel
+
+endif
+
 source "cmd/mvebu/Kconfig"
 
 config CMD_TERMINAL
cmd/Makefile +1 -0
index f1dd513a4b..06d7ad7375 100644
@@ -110,6 +110,7 @@ obj-$(CONFIG_CMD_PCI) += pci.o
 endif
 obj-$(CONFIG_CMD_PINMUX) += pinmux.o
 obj-$(CONFIG_CMD_PMC) += pmc.o
+obj-$(CONFIG_CMD_PSTORE) += pstore.o
 obj-$(CONFIG_CMD_PXE) += pxe.o pxe_utils.o
 obj-$(CONFIG_CMD_WOL) += wol.o
 obj-$(CONFIG_CMD_QFW) += qfw.o
cmd/pstore.c new file +505 -0

505 lines not on this page · raw message

doc/index.rst +7 -0
index cd98be6cc5..c556cdb607 100644
@@ -98,6 +98,13 @@ Android-specific features available in U-Boot.
 
    android/index
 
+Command line
+------------
+.. toctree::
+   :maxdepth: 2
+
+   pstore.rst
+
 Indices and tables
 ==================
 
doc/pstore.rst new file +74 -0

74 lines not on this page · raw message

Signature · 1 line
2.18.0
#2 lore raw

PATCH v3 2/3 test: Add PStore command tests

In reply to Frédéric Danis

Add PStore command to sandbox and sandbox64 defconfigs. Add test checking:
- 'pstore display' of all records
- 'pstore display' only the 2nd dump record
- 'pstore save' of all records

---
 configs/sandbox64_defconfig                |   1 +
 configs/sandbox_defconfig                  |   1 +
 test/py/tests/test_pstore.py               |  73 +++++++++++++++++++++
 test/py/tests/test_pstore_data_console.hex | Bin 0 -> 4096 bytes
 test/py/tests/test_pstore_data_panic1.hex  | Bin 0 -> 4096 bytes
 test/py/tests/test_pstore_data_panic2.hex  | Bin 0 -> 4096 bytes
 6 files changed, 75 insertions(+)
 create mode 100644 test/py/tests/test_pstore.py
 create mode 100644 test/py/tests/test_pstore_data_console.hex
 create mode 100644 test/py/tests/test_pstore_data_panic1.hex
 create mode 100644 test/py/tests/test_pstore_data_panic2.hex
configs/sandbox64_defconfig +1 -0
index 71a4d7fccb..c36cac1ffc 100644
@@ -68,6 +68,7 @@ CONFIG_CMD_CBFS=y
 CONFIG_CMD_CRAMFS=y
 CONFIG_CMD_EXT4_WRITE=y
 CONFIG_CMD_MTDPARTS=y
+CONFIG_CMD_PSTORE=y
 CONFIG_MAC_PARTITION=y
 CONFIG_AMIGA_PARTITION=y
 CONFIG_OF_CONTROL=y
configs/sandbox_defconfig +1 -0
index f96891ecae..155190a3d3 100644
@@ -77,6 +77,7 @@ CONFIG_CMD_CBFS=y
 CONFIG_CMD_CRAMFS=y
 CONFIG_CMD_EXT4_WRITE=y
 CONFIG_CMD_MTDPARTS=y
+CONFIG_CMD_PSTORE=y
 CONFIG_MAC_PARTITION=y
 CONFIG_AMIGA_PARTITION=y
 CONFIG_OF_CONTROL=y
test/py/tests/test_pstore.py new file +73 -0

73 lines not on this page · raw message

test/py/tests/test_pstore_data_console.hex new file binary +0 -0
new file mode 100644
index 0000000000000000000000000000000000000000..e7f426e8928a2793457baced5f66ee165ef429f6
GIT binary patch

zcmcgv%TnVw6lJ!(nP0e*4NyZ8zvNd{7IX(_s%e at b&{Z=OMcI}D8f?ie*$MgYGC1f z?1ZL6Cktj4lpUjc?>YA&9Sz@~d<cKp+4<+!ogJz&$oSm@K6cyiWPFg4nS8)#lU-a z1K>Db&-IDxkz1&BYW{HH_2 at lNt}hF%c=vQY{D}JqApUVz+M^#mQS38q21kR=dA1k z656*dwDxHrn#gIb!N!=1-E&>xgDwrjz_Af2FP at z4lvdzaX=YhgZ%XBT48sLX{ZLm_ zPDKnyPbK0<-l>$;Z%Z$c>pK{J at i~S|h6zy>7x$oN7_fK;cGUADG6$&=L1hs0QKS( zril``cu(&!L?=-Xw9IKpfIgDFSv1Kf*Ch~>qQYlPHMl7^+#x2DWEeiw}D?30&Lu zqsYURQS9u;kd$Sj3aJL2(beJA^{4}~WayepG3b;^j(NRa+%N$G|L&T41flTrm77 zl7bC7{aUMj%0)j*1m-ZU)vU{&ScXQN4qV at YOcuqU(?+|)pqi=961Pu$5^ROr3u5<e zapyekE1qT#SQ)0Cy>>IHSVMt?8z_X9<$KXq6bOmEx2Ld5{qrht!K;=+w26WfMvl7 z1>cRy|3%?>6 at pC^h-aB;+hPDB68mpU;l%*dCB+f#>I%E4o7RHnOt%9;Ht58$JuI- zLo0}bx8-35VtqH2im9T at M1}KRN-nA;A$J3z$nv)=j3E at hEur;=!7Ztj0?Bl_pzNtF zjlYSBYZKSnro{}C=xv$&q%>0FNuaB>)$ulQA_0lE+J8fr#J2Ueq+XO~-ehexHVu3f z+HSPfz=tG?ZTr7P-qfH4|I;(Wv3r4M2m9e%rnJu_%!+}!XbeZexIbgf{TxTf at 6nx zvRFM<JuoF$g~G at 3@z}1a1P?4aUqcn2pi;Vd-OVlR6bFp)``o9*<CZ8m#qv09<_= zTzok at +n=A!2BeRA4SVDCVuFDWhrPcU(S#B&X2B!M_=!*FEFDB&Wa)D;t{w8A&m%1 z=xpJE^glz7M?3(mi@q|u{~>SuS at 8HeUg^Ce_ZD|{ut+SirtG_j?6(l^{c0)Tij6y z4&r)XA3h$QpB#M;E at e3LOn;9qvWaiw}3mY{Vg_+p8W at WJv%&yQC*f&RWPx8;UcQv zL!CC4SPSt+yMN1Ci6o~t4smZ|I8IP7(YZdu6hQM|Fft1IlSK7%dEalo_I>6?Nk~#^ z(a>XF5~1RxV-lsrVTlz)D65$2+NS<XM^b^3D5+7P9RTjbMo6L>G+G<5XJn0kW58}f z<v3faRAa?7LPcSXT(_=<G8)R_Cnvg#kE!TFqgIM8y2*EMRiW+RmG#MY=kYEmXI zrAY_LO2E|GAK#}f*pa)|f>M{uAkHkc1BL2$tc6J8@%UR!1|D8KNeQiv=1DMOme z?6J8p=$<K&xnfv8DlsIr-W3znxEh_FTpeA0UR0_P$%I>l(?Bes(Zbaw+Qo(-P?M5 zQSeKaF&sa-bU0SBfQ!>1-YWK5n<h8=OQ$$|mKmxS3mUd7h at Vc+&Nf4IT-=slBm6N@ z6fY59~L!+U2~)PeffL$0C4-z*ic}#DRjJu82zyp-?b=YE#_C4d0#IUt@7XD2%X_ zH^h#xc$at)Y?kl#cjv>l_H-37ck}31Kv1HH*46{njYvf$=F?HcDDmvrxTgu4T3bU zgZ}O+g#jS}l74 at 6c|1a2V+niB)v+kDG=H>9bHTxz<s;0Uf^ll9jC&LlNs4pY(sdmt zFv@o=-9y}mo;O~-pwH+MitB)7rtUXWJy->2~I4A*3+b1CCYVOyobGFXTr{xV}Cau yo=L!&&H!;E#e_Fr-O-?OIiLfAGn|6heHJ!alb>%~#CVvF0Qi-|XP{eJ;km?x0{

literal 0
HcmV?d00001

test/py/tests/test_pstore_data_panic1.hex new file binary +0 -0
new file mode 100644
index 0000000000000000000000000000000000000000..988929d12c25fee4c4242776a2ed0d78cdc55948
GIT binary patch

zOWN}#Pg|hqbN__&AMD at IkJ{C)WS@hgs2028kTc&5haZRBi*LU_|MzbgbKkG=cwr<w zPlzyjkZ>U-KfAd1$7R)g?$M>HTkqai!8fmnyeN7Q${^(Yil``WO<hv6FlnXB)Rf9D zu8V~w+b+~iaYN>9Ix}UFLa}f6+H_s5(nim$>3Z6tv^7;~VdrU6l-1l4pwri7lg{kC z!Wg_bhiL)38$sG?YL=^o(aYL)1MSS#rJ6fCvNy)wxL)}t)zvDm?HkIowE3|~b4-e= zw`H2msi at LcWpmrmZQoF<c9}N;Wo@0Kds)*@O>0tP^|GvL+cX<_C*9u8R5PQW=0ygj ze6CtcwJMBEt2cyBU2EZM+Pu<hyRdZuao(i*x?Zj-QZv0t5R|)i1&n-&No+d^<1<t1 zE{oeXt>#tHDYS~LLzrnpE2Gmc9gT5a*1bBbja<#!yrfl=bvk&LMY9<%{DB at dx+&6f zX-F+pyR6dMUat@nQc=x*Ov~FQg)P=oi*%-tOg(>{BB4|ky0-HHQKToh1X0c}|1=@+ z=*h)|g2<=yWm!%LS$nC9MQhE3zNh{<?;F7bFL}Vv<e93=vLf*C!e)h9P&Xi>GhZ*m zp)=nPV*iRBtJ=I(;KV|ec0!M8Ya6Y=r{clWzeswvQs&Exm74!hV5Lqk>kF9y at b~N z5m at 4J?}{$Zsc1;G1m at IlFxcj+f>(B@;~rf!N0^$JhT%cjxpmOo^S0CJP53AkhcsL zzJ^6Gwun7nTv4TILa)BkJTjs3<8QFNi01XSQ+PIux1Ci}Pe^k2cw*=~2$0+>YAP66 zo#n<RF*fO5j3MTGQR$I&63OU_YL8E-1Lr&-suO^1MGi5Bs6Ah2$j{I^47m(o9qrY9 zsLQ<=K@4}I6cB@*$cObsN>4^Hz{GysSyD;5>E7b5#d4PJ5i3ZB4ZI?MIO!7bu*#G zU(audo;-in&~ica=*hP6bZIQT15%x5cZQ759#6ar$jI8b!itZEtU&AoNW|_E?4Kv zMO!cD;5OHxh$BN_G|I_ at gw{OZfg-wic-|*4#3XrNL}l!ODqvW>)UK*vH=I at lwgT4h z3u;=WUt at olMN3~3zm+mFvB+X at bbt_X5nj;?r#qoje!E-rb_p9NS>$G9lXyW0wiz2^ zng6#3%P-<A47Rjz+1|&LB#}rQz5 at Ox9usUE87D_?$ps2v$8#j%PoI16ehgy;Tx-5 zf at eGYh{xRP at b#Xi at 0wT~4Odo7B=5B!W=uoTI(4N&UC@R;}GAIeu7mo*7ot|yhqEO zNvLA1Y;Wr(5~?USE?T8>0a)&ZSpd%68NVNdj at l;YJA8kepWN&4eUp1TSlwsC-{W^v z%aQ+0!$&pQ=STQ)C>_2Je8lEk5BRR(T#_|X=C+GJ3~}1Kscpl?vhvVq<frfHCo;qj za7VyCH+k^k{O3s at Tj<h7_b#K4Ib@gv|qImeiC>czWil;5%@FY_!c!G$7KJQC^?A zV2B?^GQplPx#Pd)f at 9X3Xu*hn$y6Ud*s_TqwCoxp8N|QIzZ&HqwJYfG9sfIg2fG1@ zylU*wxW*M{FA-envsyqw<)1&;Ps8Hpab2?z?OVU-FnTn<shBh3oOGRipgGKDL;{ zcl_T5fXk3Gi9Og#YgCvC*N`F#B>9~Z3i4KYm-b!H at SKQ}AdtWx9dxJfAFlT)ub1Uk z%@qusho@<O{&xB}O*41?dWoxWyUS_JC2p_1hr3P3=wa>-*3-8&E(Gp|h;zQJi|cD! z!-b6DB64+i at 4<_4Q{H*_u=&8vrfN7h73#gaQKg+fc!=C2lCcND{Qy<YW$8nfL-4L- zY%+9{H#=nC#U6~2vAZJhvqQeiYy|6Shl3r0WWDO9;1dn_G6Fwp4Au=wZLJ%My7X> zMh?DF{R)ED#0dFS3$;B?|>%U4G%FrH+xTyRxS#-kGnvR`Hs;55_JK5c96Ag(_A&J zV+O8~bF7$KurL?_$D#0$*6(h$t=a-Ni|zpLkK|m%v^$aw_TI=g5ee@~wksL$Kbw7Z z_a^^^lCh&}!b22||9i>U<9kuyuDt(s$rzO5#Ft(I;+*65h3i7-u4*Wkc^Dx{0scp3 z*7!kS$O?s8jiFN;>Q3oC4Yp*0zpWBm^6c#F6<PdIY-@#Ey^6C~D6OI_;YM3dS5Uae z*6jH3Up(meat@l-6v~e?C*!{XZDyK{S(cP86LuCjrW>;f!{-7CyY(mcWnC7^&YO- z!6n&$0wcM$uT=PJUKamIeLzcpvsQ$IX-58gW8%M;?p%~&88>pfU!&Z)%y4lsCeAi z(fW{0r?@YEvSw_0#GW&vDO(dKV$fa9-)Qj at LM{XqWM{8baKe(+4Q4zCt!4qI8Oe- gK1K6im&c6#eu;b{uY4j*Km4t at E9X-&W;cM16+S2TL1t6

literal 0
HcmV?d00001

test/py/tests/test_pstore_data_panic2.hex new file binary +0 -0
new file mode 100644
index 0000000000000000000000000000000000000000..8f9d56cbe01e38f5f8fac6d46491497c9d7bda8f
GIT binary patch

zf5FMAE4S?VD|r2$5T?z9a>c=%k at v4W6_m@=ygnlbA96oGcSU*=6U(|OOlnVWwV7j z>f at Gq{AIrR=zpNs${*2dOJDYJg;H9ak)N)8Sa0)HK{i|GFAnH+2-dPpP_kW-_D!3? zSpdcGEXz2Zl7~FBc?kYjoTorSrhEoq2!0mlg3WiQfg%*J|KREPr!4;GfriiLA?mV zN3{5ucERFLnK2Ug&{V4ho~l2_&&Z;7F5Jnh+l88-A=p+J?%|Er2LWHKj~c#zceq zXj)0b+X&rq{~$Ps3NbjY^yhQG-pz9iAqEHx1_=DX$wt{5*u4jP&@KzWX at pKg=609c z3VH6<t4(42rq{4)=-N+j(K&@;ocR(0MDw_N>{Djj7Maqpl>;&#)51d^5vl+pv<at zj-J5uE4RD#JQUmV6WUxMi$TsZW);v5_^!kKD;T$vUwBysz_b=slI4_5y~P&-!Mhv zCzwHoGl8|Io14CZ(Ot)4f%^J7L~P#BGrTQk1{}-zj2xf-!KYYotMxiBKJ;>lfx#ND zHcSbU!O3h&C^=p&m$?1Pt=FME;^dvni)~1!<ZWGhGeVBrDufBM6MCo)UAbJY$qPni zVUfFnR6i(v$o)$8Z&cKB{L)snO?avkisp)l>A(Vu)*vJB%l2Dry{@aTzWLyxyxVk zZ6F7;@G<v+yu)3C$ZM=_Xx%1_uUx+?uJR(CFOXigrSqY4Sgf;P>J-?VudZTf{6&m? z3<)15mBB!@vEOs52kHXHtyL}Tmg^355lJu|CY0r9LT&H9Q$0|D#ImSB8{R!RJ&BcN zqp89VvWH^~yVBirtB<$r6Y8OYzI!@C>rz;!~hA{Y9z$HVqmhq7~tDb47lJ7Pz<vF zB?ejoDh-OkBHSSc9E4<;pxFb(0KOsy_yGek;1ZCoB7P$VxS<w<xU<uMZgIaDa0Qxm z5gJVlxRwH1K8o6OTSa<Fbq}A%l)Wtm++Z%*1_FM(!1ipV!$!6#AAp776S0r(*_<< z3<Rf|vzFEF7nIILE&WFt+Ajt|U^QdWp%EnvPTSr&>U*Dh$^sG4Ctt0ASC6Hzef=R zfg~Ke#Hw>;ql$shjLJ5t#u5Wzn6ynFH=-B_i<;k-?wDc#6o-hWgBx=)047Xy?rkxM z4LA#-nGCA8j1QC#x|Tqv9rcqn6u3j7b>;VM54F8PxMtoeSde~pN0S{Mplyty}{qt z_1DJXh)^#8={p)9mF!k-|L0|B&+eWP=HuQ5 at 7NBfseSaif)3z4cr+8qtH|IGePv zqZ2)7V6 at Hgafu!bx~|RhF}r at Sii*}H4<dS(Y0D~wEW-vTp(5X+?k%G3Ee1o7q(~$U zv+p_7k-29Wz<x&N;Rk4j1Q1TSV^%@d-6GXk|fWP<e$Hi&q?ylGxFx!r%#jQ`~Lte C8DA&>

literal 0
HcmV?d00001

Signature · 1 line
2.18.0
#3 lore raw

PATCH v3 2/3 test: Add PStore command tests

Dear Fr?d?ric,

test/py/tests/test_pstore.py line 1

Quoted 4 lines · In message <[email protected]> you wrote:

+++ b/test/py/tests/test_pstore.py

test/py/tests/test_pstore.py new file +1 -0
@@ -0,0 +1,73 @@
+# SPDX-License-Identifier: GPL-2.0

Can we please make new code always GPL-2.0+ ?

Thanks!

Best regards,

Wolfgang Denk

Signature · 8 lines
DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
That's their goal, remember, a goal that's really contrary to that of
the programmer or administrator. We just want to get our  jobs  done.
$Bill  just  wants  to  become  $$Bill. These aren't even marginallly
congruent.
         -- Tom Christiansen in <[email protected]>
#4 lore raw

PATCH v3 3/3 cmd: Fixup DT to pass PStore Ramoops parameters

In reply to Frédéric Danis

To simplify configuration and keep synchronized the PStore/Ramoops between U-Boot and the Linux kernel, this commit dynamically adds the Ramoops parameters defined in the U-Boot session to the Device Tree.

---
 cmd/pstore.c          | 38 ++++++++++++++++++++++++++++++++++++++
 common/image-fdt.c    |  4 ++++
 doc/pstore.rst        |  2 ++
 include/fdt_support.h |  3 +++
 4 files changed, 47 insertions(+)
cmd/pstore.c +38 -0
index 4e4d70d604..6cad635620 100644
@@ -479,6 +479,44 @@ static int do_pstore(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
 	return c->cmd(cmdtp, flag, argc, argv);
 }
 
+void fdt_fixup_pstore(void *blob)
+{
+	char node[32];
+	int  nodeoffset;	/* node offset from libfdt */
+
+	nodeoffset = fdt_path_offset(blob, "/");
+	if (nodeoffset < 0) {
+		/* Not found or something else bad happened. */
+		log_debug("fdt_path_offset() returned %s\n", fdt_strerror(nodeoffset));
+		return;
+	}
+
+	nodeoffset = fdt_add_subnode(blob, nodeoffset, "reserved-memory");
+	if (nodeoffset < 0) {
+		log_debug("Add 'reserved-memory' node failed: %s\n",
+				fdt_strerror(nodeoffset));
+		return;
+	}
+	fdt_appendprop_u32(blob, nodeoffset, "#address-cells", 2);
+	fdt_appendprop_u32(blob, nodeoffset, "#size-cells", 2);
+	fdt_appendprop(blob, nodeoffset, "ranges", NULL, 0);
+
+	sprintf(node, "ramoops@%llx", (unsigned long long)pstore_addr);
+	nodeoffset = fdt_add_subnode(blob, nodeoffset, node);
+	if (nodeoffset < 0) {
+		log_debug("Add '%s' node failed: %s\n", node, fdt_strerror(nodeoffset));
+		return;
+	}
+	fdt_appendprop_string(blob, nodeoffset, "compatible", "ramoops");
+	fdt_appendprop_u64(blob, nodeoffset, "reg", pstore_addr);
+	fdt_appendprop_u64(blob, nodeoffset, "reg", pstore_length);
+	fdt_appendprop_u32(blob, nodeoffset, "record-size", pstore_record_size);
+	fdt_appendprop_u32(blob, nodeoffset, "console-size", pstore_console_size);
+	fdt_appendprop_u32(blob, nodeoffset, "ftrace-size", pstore_ftrace_size);
+	fdt_appendprop_u32(blob, nodeoffset, "pmsg-size", pstore_pmsg_size);
+	fdt_appendprop_u32(blob, nodeoffset, "ecc-size", pstore_ecc_size);
+}
+
 U_BOOT_CMD(pstore, 10, 0, do_pstore,
 	   "Manage Linux Persistent Storage",
 	   "set <addr> <len> [record-size] [console-size] [ftrace-size] [pmsg_size] [ecc-size]\n"
common/image-fdt.c +4 -0
index 3002948b6b..491d55ad1a 100644
@@ -547,6 +547,10 @@ int image_setup_libfdt(bootm_headers_t *images, void *blob,
 	}
 	/* Update ethernet nodes */
 	fdt_fixup_ethernet(blob);
+#if CONFIG_IS_ENABLED(CMD_PSTORE)
+	/* Append PStore configuration */
+	fdt_fixup_pstore(blob);
+#endif
 	if (IMAGE_OF_BOARD_SETUP) {
 		fdt_ret = ft_board_setup(blob, gd->bd);
 		if (fdt_ret) {
doc/pstore.rst +2 -0
index 3ed2e9e0fd..3c84b8b172 100644
@@ -24,6 +24,8 @@ i.e.::
 
 The same values should be set in U-Boot to be able to retrieve the records.
 This values can be set at build time in U-Boot configuration file, or at runtime.
+U-Boot automatically patches the Device Tree to pass the Ramoops parameters to
+the kernel.
 
 The PStore configuration parameters are:
 
include/fdt_support.h +3 -0
index ba14acd7f6..7afbdcfe37 100644
@@ -350,4 +350,7 @@ int fdt_update_ethernet_dt(void *blob);
 #ifdef CONFIG_FSL_MC_ENET
 void fdt_fixup_board_enet(void *blob);
 #endif
+#ifdef CONFIG_CMD_PSTORE
+void fdt_fixup_pstore(void *blob);
+#endif
 #endif /* ifndef __FDT_SUPPORT_H */
Signature · 1 line
2.18.0
#5 lore raw

PATCH v3 3/3 cmd: Fixup DT to pass PStore Ramoops parameters

Quoted 19 lines · Fr?d?ric Danis

To simplify configuration and keep synchronized the PStore/Ramoops between U-Boot and the Linux kernel, this commit dynamically adds the Ramoops parameters defined in the U-Boot session to the Device Tree.

---
 cmd/pstore.c          | 38 ++++++++++++++++++++++++++++++++++++++
 common/image-fdt.c    |  4 ++++
 doc/pstore.rst        |  2 ++
 include/fdt_support.h |  3 +++
 4 files changed, 47 insertions(+)
cmd/pstore.c +0 -0
index 4e4d70d604..6cad635620 100644
@@ -479,6 +479,44 @@ static int do_pstore(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
 	return c->cmd(cmdtp, flag, argc, argv);
 }

cmd/pstore.c static int do_pstore(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 490

Quoted 10 lines
cmd/pstore.c +9 -0
 
+void fdt_fixup_pstore(void *blob)
+{
+	char node[32];
+	int  nodeoffset;	/* node offset from libfdt */
+
+	nodeoffset = fdt_path_offset(blob, "/");
+	if (nodeoffset < 0) {
+		/* Not found or something else bad happened. */
+		log_debug("fdt_path_offset() returned %s\n", fdt_strerror(nodeoffset));

Thanks for implementing device tree support.

Should this be log_err()?

cmd/pstore.c static int do_pstore(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 500

Quoted 10 lines
cmd/pstore.c +10 -0
+		return;
+	}
+
+	nodeoffset = fdt_add_subnode(blob, nodeoffset, "reserved-memory");
+	if (nodeoffset < 0) {
+		log_debug("Add 'reserved-memory' node failed: %s\n",
+				fdt_strerror(nodeoffset));
+		return;
+	}
+	fdt_appendprop_u32(blob, nodeoffset, "#address-cells", 2);

For adding the first value I guess it is preferable to use the fdt_set* functions.

fdt_setprop_u32()

cmd/pstore.c static int do_pstore(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 501

Quoted 1 line
cmd/pstore.c +1 -0
+	fdt_appendprop_u32(blob, nodeoffset, "#size-cells", 2);

fdt_setprop_u32()

cmd/pstore.c static int do_pstore(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 502

Quoted 1 line
cmd/pstore.c +1 -0
+	fdt_appendprop(blob, nodeoffset, "ranges", NULL, 0);

fdt_setprop_empty()?

cmd/pstore.c static int do_pstore(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 507

Quoted 5 lines
cmd/pstore.c +5 -0
+
+	sprintf(node, "ramoops@%llx", (unsigned long long)pstore_addr);
+	nodeoffset = fdt_add_subnode(blob, nodeoffset, node);
+	if (nodeoffset < 0) {
+		log_debug("Add '%s' node failed: %s\n", node, fdt_strerror(nodeoffset));

Should this be log_err()?

cmd/pstore.c static int do_pstore(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 510

Quoted 3 lines
cmd/pstore.c +3 -0
+		return;
+	}
+	fdt_appendprop_string(blob, nodeoffset, "compatible", "ramoops");

fdt_setprop_string()

cmd/pstore.c static int do_pstore(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 511

Quoted 1 line
cmd/pstore.c +1 -0
+	fdt_appendprop_u64(blob, nodeoffset, "reg", pstore_addr);

fdt_setprop_u64()

cmd/pstore.c static int do_pstore(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 512

Quoted 1 line
cmd/pstore.c +1 -0
+	fdt_appendprop_u64(blob, nodeoffset, "reg", pstore_length);

This one is ok.

cmd/pstore.c static int do_pstore(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) · line 513

Quoted 1 line
cmd/pstore.c +1 -0
+	fdt_appendprop_u32(blob, nodeoffset, "record-size", pstore_record_size);

fdt_setprop_u32(), same below.

Best regards

Heinrich

Quoted 50 lines
cmd/pstore.c +6 -0
+	fdt_appendprop_u32(blob, nodeoffset, "console-size", pstore_console_size);
+	fdt_appendprop_u32(blob, nodeoffset, "ftrace-size", pstore_ftrace_size);
+	fdt_appendprop_u32(blob, nodeoffset, "pmsg-size", pstore_pmsg_size);
+	fdt_appendprop_u32(blob, nodeoffset, "ecc-size", pstore_ecc_size);
+}
+
 U_BOOT_CMD(pstore, 10, 0, do_pstore,
 	   "Manage Linux Persistent Storage",
 	   "set <addr> <len> [record-size] [console-size] [ftrace-size] [pmsg_size] [ecc-size]\n"
common/image-fdt.c +4 -0
index 3002948b6b..491d55ad1a 100644
@@ -547,6 +547,10 @@ int image_setup_libfdt(bootm_headers_t *images, void *blob,
 	}
 	/* Update ethernet nodes */
 	fdt_fixup_ethernet(blob);
+#if CONFIG_IS_ENABLED(CMD_PSTORE)
+	/* Append PStore configuration */
+	fdt_fixup_pstore(blob);
+#endif
 	if (IMAGE_OF_BOARD_SETUP) {
 		fdt_ret = ft_board_setup(blob, gd->bd);
 		if (fdt_ret) {
doc/pstore.rst +2 -0
index 3ed2e9e0fd..3c84b8b172 100644
@@ -24,6 +24,8 @@ i.e.::
 
 The same values should be set in U-Boot to be able to retrieve the records.
 This values can be set at build time in U-Boot configuration file, or at runtime.
+U-Boot automatically patches the Device Tree to pass the Ramoops parameters to
+the kernel.
 
 The PStore configuration parameters are:
 
include/fdt_support.h +3 -0
index ba14acd7f6..7afbdcfe37 100644
@@ -350,4 +350,7 @@ int fdt_update_ethernet_dt(void *blob);
 #ifdef CONFIG_FSL_MC_ENET
 void fdt_fixup_board_enet(void *blob);
 #endif
+#ifdef CONFIG_CMD_PSTORE
+void fdt_fixup_pstore(void *blob);
+#endif
 #endif /* ifndef __FDT_SUPPORT_H */
#6 lore raw

v3 cover Add command to display or save Linux PStore dumps

In reply to Frédéric Danis

Dear Fr?d?ric,

Quoted 8 lines · In message <[email protected]> you wrote:

This serie of patches adds a new pstore command allowing to display or save ramoops logs (oops, panic, console, ftrace and user) generated by a previous kernel crash.
PStore parameters can be set in U-Boot configuration file, or at run-time using "pstore set" command. For kernel using Device Tree, the parameters are dynamically added to Device Tree.
Records size should be the same as the ones used by kernel, and should be a power of 2.

I wonder if we are reinventing the wheel here again?

There is this feature in U-Boot which is called "shared log buffer"; a couple of years ago this was fully functional at least on Power and ARM architectures, but it was rarely used and probably has not been tested for years. A;so, the necessary tiny patch to have it supported in Linux as well never made it upstream (don't remember why, likely lack of time/interest).

The functionality we had then was the following:

- A memory area war reserved in U-Boot (typically at the upper end of memory) as a buffer that was shared between U-Boot and Linux. The format was as used for the kernel log buffer.
- Upon boot, U-Boot would not re-initialize an existing log buffer, but keep it's content. That means, you could read and display the log buffer of the linux kernel that was running before the reset. After kernel crashes, pretty often this contained information that the kernel could not even print to the serial console any more.
- In U-Boot, you could append log entries to that buffer. For example, this was used to record the results of the Power On Self Test (POST) routines (another feature that only few people still remember).
- When booting Linux, the kernel syslog mechanism was used to extract the information from the log buffer in the usual way.

The interesting fact here was that the Linux kernel was able to extract and save the kernel panic messages etc. from the crash before, plus any messages logged by U-Boot.

To me this sounds very much like what you are adding here (plus a few features more). Does it make sense to unify such code?

Added Heiko to Cc:, as he is currently working on fixes to get shared logbuffer working again for another project.

Best regards,

Wolfgang Denk

Signature · 5 lines
DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
I'm what passes for a Unix guru in my office. This is  a  frightening
concept. - Lee Ann Goldstein, in <[email protected]>
#7 lore raw

v3 cover Add command to display or save Linux PStore dumps

Quoted 38 lines · Wolfgang Denk

Dear Fr?d?ric,

Quoted 8 lines · In message <[email protected]> you wrote:

This serie of patches adds a new pstore command allowing to display or save ramoops logs (oops, panic, console, ftrace and user) generated by a previous kernel crash.
PStore parameters can be set in U-Boot configuration file, or at run-time using "pstore set" command. For kernel using Device Tree, the parameters are dynamically added to Device Tree.
Records size should be the same as the ones used by kernel, and should be a power of 2.

I wonder if we are reinventing the wheel here again?

There is this feature in U-Boot which is called "shared log buffer"; a couple of years ago this was fully functional at least on Power and ARM architectures, but it was rarely used and probably has not been tested for years. A;so, the necessary tiny patch to have it supported in Linux as well never made it upstream (don't remember why, likely lack of time/interest).

The functionality we had then was the following:

- A memory area war reserved in U-Boot (typically at the upper end of memory) as a buffer that was shared between U-Boot and Linux. The format was as used for the kernel log buffer.
- Upon boot, U-Boot would not re-initialize an existing log buffer, but keep it's content. That means, you could read and display the log buffer of the linux kernel that was running before the reset. After kernel crashes, pretty often this contained information that the kernel could not even print to the serial console any more.
- In U-Boot, you could append log entries to that buffer. For example, this was used to record the results of the Power On Self Test (POST) routines (another feature that only few people still remember).
- When booting Linux, the kernel syslog mechanism was used to extract the information from the log buffer in the usual way.

Quoted 7 lines

The interesting fact here was that the Linux kernel was able to extract and save the kernel panic messages etc. from the crash before, plus any messages logged by U-Boot.

To me this sounds very much like what you are adding here (plus a few features more). Does it make sense to unify such code?

It seems you are relating to
https://lore.kernel.org/lkml/844oyrqvvb.fsf at sauna.l.org/t/

ramoops in Linux is exactly doing what was suggested in 2009. You can find the Documentation/admin-guide/ramoops.rst

git grep -GHrn 'shared log' finds nothing in U-Boot. So if any part of the old implementation in U-Boot exists, could you, please, point us to the coding.

If the original design never made it into Linux and there is an established Linux interface since 2011, I would plead to eliminate any remaining non-compliant coding from U-Boot should it exist.

Best regards

Heinrich

Quoted 10 lines

Added Heiko to Cc:, as he is currently working on fixes to get shared logbuffer working again for another project.

Best regards,

Wolfgang Denk

#8 lore raw

v3 cover Add command to display or save Linux PStore dumps

Dear Heinrich,

Quoted 6 lines · In message <[email protected]> you wrote:
Quoted 3 lines

To me this sounds very much like what you are adding here (plus a few features more). Does it make sense to unify such code?

It seems you are relating to
https://lore.kernel.org/lkml/844oyrqvvb.fsf at sauna.l.org/t/

No, I'm not. I was talking of my own code from many, many years ago.

Quoted 2 lines

ramoops in Linux is exactly doing what was suggested in 2009. You can find the Documentation/admin-guide/ramoops.rst

We had this in U-Boot long before that time. It was a key requirement when we added POST support in 2002.

Quoted 3 lines

git grep -GHrn 'shared log' finds nothing in U-Boot. So if any part of the old implementation in U-Boot exists, could you, please, point us to the coding.

The shared log buffer support was added by commit 56f94be3ef63:

commit 56f94be3ef63732384063e110277ed89701b6471 (tag: LABEL_2002_11_05_1735)

Author: Wolfgang Denk <wdenk>
Date:   Tue Nov 5 16:35:14 2002 +0000

* Add support for log buffer which can be passed to Linux kernel's syslog mechanism; used especially for POST results.

* Patch by Klaus Heydeck, 31 Oct 2002:
Add initial support for kup4k board

The code was mainly in common/cmd_log.c, but this got heahily rewritten and is now in cmd/log.c ; apparently this got lost (like the original copyright, sic!) when Simon modified / rewrote this driver.

For history, try: git log --follow -- common/cmd_log.c

Quoted 3 lines

If the original design never made it into Linux and there is an established Linux interface since 2011, I would plead to eliminate any remaining non-compliant coding from U-Boot should it exist.

I understand what Linus has is one-way, only focussing on crash dump, i. e. it does not allow to pass information from U-Boot to Linux?

Also, my understanding is that the changes needed in Linux are pretty small.

Maybe Heiko can comment on that...

Best regards,

Wolfgang Denk

Signature · 4 lines
DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Ken Dodd?s dad?s dog?s dead.
#9 lore raw

v3 cover Add command to display or save Linux PStore dumps

Hello Wolfgang, Heinrich,

Am 20.03.2020 um 09:08 schrieb Wolfgang Denk:

Quoted 47 lines

Dear Heinrich,

Quoted 6 lines · In message <[email protected]> you wrote:
Quoted 3 lines

To me this sounds very much like what you are adding here (plus a few features more). Does it make sense to unify such code?

It seems you are relating to
https://lore.kernel.org/lkml/844oyrqvvb.fsf at sauna.l.org/t/

No, I'm not. I was talking of my own code from many, many years ago.

Quoted 2 lines

ramoops in Linux is exactly doing what was suggested in 2009. You can find the Documentation/admin-guide/ramoops.rst

We had this in U-Boot long before that time. It was a key requirement when we added POST support in 2002.

Quoted 3 lines

git grep -GHrn 'shared log' finds nothing in U-Boot. So if any part of the old implementation in U-Boot exists, could you, please, point us to the coding.

The shared log buffer support was added by commit 56f94be3ef63:

commit 56f94be3ef63732384063e110277ed89701b6471 (tag: LABEL_2002_11_05_1735)

Author: Wolfgang Denk <wdenk>
Date:   Tue Nov 5 16:35:14 2002 +0000

* Add support for log buffer which can be passed to Linux kernel's syslog mechanism; used especially for POST results.

* Patch by Klaus Heydeck, 31 Oct 2002:
Add initial support for kup4k board

The code was mainly in common/cmd_log.c, but this got heahily rewritten and is now in cmd/log.c ; apparently this got lost (like the original copyright, sic!) when Simon modified / rewrote this driver.

For history, try: git log --follow -- common/cmd_log.c

Quoted 3 lines

If the original design never made it into Linux and there is an established Linux interface since 2011, I would plead to eliminate any remaining non-compliant coding from U-Boot should it exist.

Quoted 7 lines

I understand what Linus has is one-way, only focussing on crash dump, i. e. it does not allow to pass information from U-Boot to Linux?

Also, my understanding is that the changes needed in Linux are pretty small.

Maybe Heiko can comment on that...

Since 2002 the logbuffer support in linux changed, so the old patch does not work anymore.

First there is now the struct "printk_log" [1]

Unfortunately the size of it is dependend on linux configuration.

Each log entry is prepended from such a struct, just missing the possibility to detect if entry is valid (as linux was the only user of it, this was not needed).

There is already a possibilty to set a log buffer length through kernel commandline parameter "log_buf_len" [2].
If this is passed, kernel already uses a new area (currently memblock_alloced) for log buffer storage!

So added now the following to linux logbuffer handling:

Pass the persistent logbuffer address and length through a kernel commandline parameter and use it instead memblock_alloc()

Add a magic field in [1] which I use to detect if there are valid entries and where is the end of logbuffer entries in the new logbuffer area (Not that easy as there was may a wrap around, but also this is detectable)

-> start and end in the new log area is now known.

Now copy the new logbuffer messages from __log_buf with the already existing functions in linux kernel to the end of the new logbuffer area ... and you have old linux bootlogs and the new ones after a reboot.

aditionally changes in U-Boot:

- enable PRAM to reserve RAM at the end of RAM for logbuffer so U-Boot does not use this memory space. Already in U-Boot.

- add common/log_buffer.c log backend, which uses the functions from kernel for analysing/writing content of the logbuffer area.

"log rec" U-Boot command appends now also U-Boot log messages to the logbuffer area.

- add "log show" command, with which you can show current content of logbuffer area

And now U-Boot log messages also shown in linux bootlog ...

bye
Heiko
[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/kernel/printk/printk.c#n368

[2] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/kernel/printk/printk.c#n1114

Signature · 3 lines
DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: +49-8142-66989-52   Fax: +49-8142-66989-80   Email: hs at denx.de
#10 lore raw

v3 cover Add command to display or save Linux PStore dumps

In reply to Wolfgang Denk

Hi Wolfgang, Heinrich, Heiko,

Quoted 35 lines · Wolfgang Denk

Dear Heinrich,

Quoted 4 lines · In message<[email protected]> you wrote:
Quoted 2 lines

To me this sounds very much like what you are adding here (plus a few features more). Does it make sense to unify such code?

It seems you are relating to
https://lore.kernel.org/lkml/844oyrqvvb.fsf at sauna.l.org/t/

No, I'm not. I was talking of my own code from many, many years ago.

Quoted 2 lines

ramoops in Linux is exactly doing what was suggested in 2009. You can find the Documentation/admin-guide/ramoops.rst

We had this in U-Boot long before that time. It was a key requirement when we added POST support in 2002.

Quoted 3 lines

git grep -GHrn 'shared log' finds nothing in U-Boot. So if any part of the old implementation in U-Boot exists, could you, please, point us to the coding.

The shared log buffer support was added by commit 56f94be3ef63:

commit 56f94be3ef63732384063e110277ed89701b6471 (tag: LABEL_2002_11_05_1735)

Author: Wolfgang Denk <wdenk>
Date:   Tue Nov 5 16:35:14 2002 +0000

* Add support for log buffer which can be passed to Linux kernel's syslog mechanism; used especially for POST results.

* Patch by Klaus Heydeck, 31 Oct 2002:
Add initial support for kup4k board

The code was mainly in common/cmd_log.c, but this got heahily rewritten and is now in cmd/log.c ; apparently this got lost (like the original copyright, sic!) when Simon modified / rewrote this driver.

Quoted 8 lines

For history, try: git log --follow -- common/cmd_log.c

Quoted 3 lines

If the original design never made it into Linux and there is an established Linux interface since 2011, I would plead to eliminate any remaining non-compliant coding from U-Boot should it exist.

I understand what Linus has is one-way, only focussing on crash dump, i. e. it does not allow to pass information from U-Boot to Linux?

Yes, currently it is not possible to pass information from U-Boot to Linux. But, PStore does not only store crash dump, but also console, ftrace and user space logs.

It may be possible to add a "bootloader" storage space to PStore.

Best regards,

Fr?d?ric Danis

#11 lore raw

v3 cover Add command to display or save Linux PStore dumps

Dear Fr?d?ric,

Quoted 9 lines · In message <[email protected]> you wrote:
Quoted 3 lines

I understand what Linus has is one-way, only focussing on crash dump, i. e. it does not allow to pass information from U-Boot to Linux?

Yes, currently it is not possible to pass information from U-Boot to Linux. But, PStore does not only store crash dump, but also console, ftrace and user space logs.

It may be possible to add a "bootloader" storage space to PStore.

This does not make much sense to me. We should not try to push any possible functionality into U-Boot just because we can.

U-Boot is a boot loader, and should only contain necessary functionality and things, that are impossible or difficult to solve in other ways.

I would much rather see the opposite direction working (again):
being able to pass information from U-Boot to Linux kernel and further into Linux user land in a standard way.

Best regards,

Wolfgang Denk

Signature · 4 lines
DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Panic, n.: The second time you can't do it the first time.
#12 lore raw

v3 cover Add command to display or save Linux PStore dumps

Dear Wolfgang,

Quoted 5 lines · Wolfgang Denk

Dear Fr?d?ric,

Quoted 2 lines · In message <[email protected]> you wrote:
Quoted 2 lines

I understand what Linus has is one-way, only focussing on crash dump, i. e. it does not allow to pass information from U-Boot to Linux?

Quoted 7 lines
Quoted 5 lines

Yes, currently it is not possible to pass information from U-Boot to Linux. But, PStore does not only store crash dump, but also console, ftrace and user space logs.

It may be possible to add a "bootloader" storage space to PStore.

This does not make much sense to me. We should not try to push any possible functionality into U-Boot just because we can.

Yes, I understand it.
But, when debugging kernel crashes occurring either early in boot or hard crashes later, it can be helpful to be able to retrieve logs using the U-Boot session.

Quoted 5 lines

U-Boot is a boot loader, and should only contain necessary functionality and things, that are impossible or difficult to solve in other ways.

Quoted 8 lines

I would much rather see the opposite direction working (again):
being able to pass information from U-Boot to Linux kernel and further into Linux user land in a standard way.

Best regards,

Wolfgang Denk

Best regards,

Fr?d?ric Danis

#13 lore raw

v3 cover Add command to display or save Linux PStore dumps

Dear Fr?d?ric,

Quoted 4 lines · In message <[email protected]> you wrote:

But, when debugging kernel crashes occurring either early in boot or hard crashes later, it can be helpful to be able to retrieve logs using the U-Boot session.

Full agreement, and we had such a feature working and in mainline already 17+ years ago.

It's sad that people rather reinvent the wheel instead of using existing functionality (or reviving it, if it has been broken over time).

Best regards,

Wolfgang Denk

Signature · 5 lines
DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Be wiser than other people if you can, but do not tell them so.
                                       -- Philip Earl of Chesterfield
#14 lore raw

v3 cover Add command to display or save Linux PStore dumps

Hi,

Quoted 9 lines · Wolfgang Denk

Dear Fr?d?ric,

Quoted 4 lines · In message <[email protected]> you wrote:

But, when debugging kernel crashes occurring either early in boot or hard crashes later, it can be helpful to be able to retrieve logs using the U-Boot session.

Quoted 6 lines

Full agreement, and we had such a feature working and in mainline already 17+ years ago.

It's sad that people rather reinvent the wheel instead of using existing functionality (or reviving it, if it has been broken over time).

Indeed.

Let's make sure there are tests added to this new feature (which looks extremely useful to me), so it stays working.

Regards,
Simon

#15 lore raw

v3 cover Add command to display or save Linux PStore dumps

Quoted 15 lines · Simon Glass

Hi,

Quoted 12 lines · Wolfgang Denk

Dear Fr?d?ric,

Quoted 4 lines · In message <[email protected]> you wrote:

But, when debugging kernel crashes occurring either early in boot or hard crashes later, it can be helpful to be able to retrieve logs using the U-Boot session.

Full agreement, and we had such a feature working and in mainline already 17+ years ago.

Quoted 8 lines
Quoted 3 lines

It's sad that people rather reinvent the wheel instead of using existing functionality (or reviving it, if it has been broken over time).

Indeed.

Let's make sure there are tests added to this new feature (which looks extremely useful to me), so it stays working.

I'll just chime in to say that I have now reached the point in my career where I've both been told "you're re-inventing the wheel (again)" and now seen the wheel be re-re...-re-invented.

Signature · 8 lines
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200323/1b2573d1/attachment.sig>