STB: primary CPU thread not aborting immedaitely after a secondary CPU job got abort. #148
Comments
|
Primary main thread took additional 6 seconds to abort the boot process, after secondary job gets abort. |
|
Ouch. We should probably think about a better way to deal with that. |
|
Adding cross reference -> https://bugzilla.linux.ibm.com/show_bug.cgi?id=165960 |
ruscur
pushed a commit
to ruscur/skiboot
that referenced
this issue
Feb 28, 2020
Here is a proposal to collect OPAL call statistics, counts and duration, and track areas we could possibly improve. With a small Linux driver to dump the stats in debugfs, here is what we get on a P9 after boot: OPAL_CONSOLE_WRITE : #22318 0/0/47 OPAL_RTC_READ : open-power#9 0/4/15 OPAL_READ_NVRAM : #3468 0/0/6 OPAL_HANDLE_INTERRUPT : #4724 0/57/10026 OPAL_POLL_EVENTS : #508 2/141/10033 OPAL_PCI_CONFIG_READ_BYTE : #3623 0/0/4 OPAL_PCI_CONFIG_READ_HALF_WORD : #5579 0/0/8 OPAL_PCI_CONFIG_READ_WORD : #6156 0/0/7 OPAL_PCI_CONFIG_WRITE_BYTE : open-power#2 0/0/0 OPAL_PCI_CONFIG_WRITE_HALF_WORD : #1282 0/0/1 OPAL_PCI_CONFIG_WRITE_WORD : #1335 0/0/1 OPAL_PCI_EEH_FREEZE_STATUS : #11123 0/0/2 OPAL_CONSOLE_WRITE_BUFFER_SPACE : #139088 0/0/11 OPAL_PCI_EEH_FREEZE_CLEAR : open-power#148 1/2/8 OPAL_PCI_PHB_MMIO_ENABLE : open-power#22 0/0/0 OPAL_PCI_SET_PHB_MEM_WINDOW : open-power#22 0/0/1 OPAL_PCI_MAP_PE_MMIO_WINDOW : open-power#56 0/0/0 OPAL_PCI_SET_PE : open-power#44 279/284/293 OPAL_PCI_SET_PELTV : open-power#66 0/0/0 OPAL_PCI_SET_XIVE_PE : #1123 0/0/1 OPAL_GET_MSI_64 : #1120 0/0/0 OPAL_START_CPU : open-power#238 8/21/35 OPAL_QUERY_CPU_STATUS : #357 0/11/69 OPAL_PCI_MAP_PE_DMA_WINDOW : open-power#16 0/0/1 OPAL_PCI_MAP_PE_DMA_WINDOW_REAL : open-power#16 0/0/1 OPAL_PCI_RESET : open-power#35 0/471/851 OPAL62 : open-power#6 0/10/46 OPAL_XSCOM_READ : open-power#26 0/0/2 OPAL_XSCOM_WRITE : open-power#8 0/0/1 OPAL_REINIT_CPUS : open-power#4 348/8247/11061 OPAL_CHECK_TOKEN : #134112 0/0/0 OPAL_GET_MSG : open-power#30 0/0/1 OPAL87 : open-power#1 0/0/0 OPAL_PCI_SET_PHB_CAPI_MODE : open-power#2 0/60/121 OPAL_SLW_SET_REG : #1080 3/3/13 OPAL_IPMI_SEND : open-power#53 0/5/11 OPAL_IPMI_RECV : open-power#53 0/0/2 OPAL_I2C_REQUEST : open-power#20 6/10/19 OPAL_FLASH_READ : open-power#10 19/10452/58305 OPAL_PRD_MSG : open-power#1 0/3/3 OPAL_CONSOLE_FLUSH : #134079 0/0/12 OPAL_PCI_GET_PRESENCE_STATE : open-power#7 1/1/3 OPAL_PCI_GET_POWER_STATE : open-power#9 0/0/0 OPAL_PCI_TCE_KILL : open-power#20 1/8/133 OPAL_NMMU_SET_PTCR : open-power#3 253/255/257 OPAL_XIVE_RESET : open-power#3 0/114709/115403 OPAL_XIVE_GET_IRQ_INFO : #1427 0/0/6 OPAL_XIVE_SET_IRQ_CONFIG : #1113 0/125/2810 OPAL_XIVE_GET_QUEUE_INFO : open-power#240 0/0/2 OPAL_XIVE_SET_QUEUE_INFO : #360 0/60/1216 OPAL_XIVE_ALLOCATE_VP_BLOCK : open-power#2 0/59/60 OPAL_XIVE_GET_VP_INFO : open-power#240 0/0/0 OPAL_XIVE_SET_VP_INFO : #360 0/298/3080 OPAL_XIVE_ALLOCATE_IRQ : open-power#240 0/0/3 OPAL140 : open-power#119 0/253/1109 OPAL_IMC_COUNTERS_INIT : open-power#60 9/10/20 OPAL_IMC_COUNTERS_STOP : open-power#36 0/0/2 OPAL_PCI_GET_PBCQ_TUNNEL_BAR : open-power#1 2/2/2 OPAL_PCI_SET_PBCQ_TUNNEL_BAR : open-power#1 1/1/1 OPAL_NX_COPROC_INIT : open-power#2 3/4/5 Signed-off-by: Cédric Le Goater <clg@kaod.org>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
If there are multiple flash resources pre-load requests came, before one CPU job gets abort the other CPU can download the PNOR resource. @cclaudio @stewart-ibm Please have a look at it.
The text was updated successfully, but these errors were encountered: