Wednesday November 08, 2006
Wednesday November 08, 2006 New misc module: pcie(7d) for x64 Solaris
PCI Express x64 Install errors: post Solaris Express Build 53
In case Solaris Express installation fails or hangs on your PCI Express system, then discard the previous suggested workaround per x64_solaris_installation_could_fail. Solaris Express Build 53 introduces a new module for x64 PCI Express Error handling called - pcie(7d). This is same as on the SPARC based machines. The workaround is now:
Step 1: Boot with -kd optionA more permanent solution is in the works.
Do break into the GRUB menu editor and append "-kd" to
the line that begins with "kernel ..."
When Solaris boots, it will pause at the kernel debugger prompt.
Step 2: Set a kernel breakpoint for the symbol "attach_drivers" thus:
[0]> ::bp -d -n 1 attach_drivers
[0]> :c
Step 3: Disable PCI Express Error Handling thus:
[0]> pcie`pcie_error_disable_flag/W 1
[0]> :c
Step 4: To make this setting permanent, append to /etc/system
set pcie:pcie_errror_disable_flag = 1
Posted by anish ( Nov 08 2006, 09:22:31 AM PST ) Permalink
Thursday June 15, 2006 Will keep updating this link with more regional tid bits. Posted by anish ( Jun 15 2006, 09:27:03 AM PDT ) Permalink
Wednesday June 14, 2006 Yes, OpenSolaris is a Gemini !!
It is probably more famous than these Geminis were when they were a year old.
Happy Birthday !!
Posted by anish ( Jun 14 2006, 09:11:48 AM PDT ) Permalink
Monday June 12, 2006 ::softint added for x64 servers
On SPARC based systems, ::softint MDB macro provided software interrupt information. This macro was not ported over to the x64 side. Now, starting with Solaris 11 Build 42 onwards one can get live software interrupt info on the x64 systems.
# echo ::softint | mdb -k
ADDR PEND PIL ARG1 ARG2 ISR(s)
ffffffff8634d880 0 1 ffffffff8c547000 0 nge_chip_factotum
ffffffff854e6800 0 1 ffffffff8c547000 0 nge_reschedule
ffffffff84ceb380 0 1 ffffffff86a0d000 0 nge_chip_factotum
ffffffff84ceb500 0 1 ffffffff86a0d000 0 nge_reschedule
ffffffff84536f00 0 1 ffffffff82bb7b40 0 errorq_intr
fffffffffbc05b08 0 1 0 0 softlevel1
ffffffff8453c380 0 2 ffffffff82bb2400 0 errorq_intr
ffffffff8453c200 0 2 ffffffff82bb7e00 0 errorq_intr
fffffffffbc008a8 0 2 0 0 cbe_low_level
ffffffff8b487d80 0 4 ffffffff91113200 0 power_soft_intr
ffffffff861db180 0 4 ffffffff856bb198 0 ghd_doneq_process
ffffffff854e6880 0 4 ffffffff856bb198 0 ghd_timeout_softintr
ffffffff8453c800 0 4 0 0 asysoftintr
ffffffff84536d80 0 4 ffffffff842248d8 0 ghd_doneq_process
ffffffff84536c80 0 4 ffffffff842248d8 0 ghd_timeout_softintr
ffffffff8453cf80 0 9 ffffffff8558eda0 0 hcdi_soft_intr
ffffffff8453cb00 0 9 ffffffff82cd88c0 0 hcdi_soft_intr
fffffffffbc00868 0 10 0 0 cbe_softclock
#
Note that with Advanced DDI Interrupt Interfaces, interrupt service routines take two arguments (ARG1, ARG2) and the software interrupt identifier is ADDR.
Technorati Tag: OpenSolaris Technorati Tag: Solaris Posted by anish ( Jun 12 2006, 12:11:26 PM PDT ) Permalink
Friday June 09, 2006 x64: Solaris Installation could fail with certain PCI Express chipsets
Solaris Express x64 Installation failures with certain PCI Express Chipsets
Solaris Express x64 installation, after PCI Express support was added, could fail with certain
PCI Express chipsets. This is specially seen with ATI's RS480 chipsets. It has been fixed in
Solaris Express Build 41. Interim, if you run into this, you could apply this workaround:
Step 1: Boot with -kd optionNote that, above is not needed post Solaris Express Build 41.
You could break into GRUB menu editor
and append "-kd" to the line that begins with "kernel ..."
When Solaris boots, it will stop at the kernel debugger.
Step 2: Set a debugger breakpoint for the symbol "attach_drivers" thus:
::bp -d -n 1 attach_drivers
:c
Solaris will hit above breakpoint
Step 3: Disable Error Handling
npe`pcie_error_disable_flag/W 1
:c
Step 4: This setting could be made permanent by adding this line to /etc/system
set npe:pcie_errror_disable_flag = 1
Technorati Tag: OpenSolaris Technorati Tag: Solaris Posted by anish ( Jun 09 2006, 09:41:46 AM PDT ) Permalink
Wednesday April 26, 2006 Fun stuff on the road
Driving to work, noticed a car identified itself as License Plate Recognition Vehicle using PIPS Technology.
Checked out their website and it was kind cool to see how license plates are being read and verified.
Posted by anish ( Apr 26 2006, 04:53:04 PM PDT ) Permalink
Friday April 14, 2006 Equivalent of 'cat /proc/interrupts'
New MDB dcmd ::interrupts
Solaris Express now has support for a new MDB dcmd ::interrupts
that shows similar output to that of 'cat /proc/interrupts' on Linux.
Output of ::interrupts on a x64 running 64-bit Solaris Express:
(System is using IOAPIC)
# echo ::interrupts | mdb -k IRQ Vector IPL Bus Type CPU Share APIC/INT# ISR(s) 3 0xb1 12 ISA Fixed 1 1 0x0/0x3 asyintr 4 0xb0 12 ISA Fixed 1 1 0x0/0x4 asyintr 6 0x42 5 ISA Fixed 3 1 0x0/0x6 fdc_intr 9 0x80 9 PCI Fixed 1 1 0x0/0x9 acpi_wrapper_isr 15 0x41 5 ISA Fixed 2 1 0x0/0xf ata_intr 16 0x81 9 PCI Fixed 1 3 0x0/0x10 hci1394_isr, uhci_intr, uhci_intr 17 0x82 9 PCI Fixed 2 1 0x0/0x11 audio810_intr 18 0x40 5 PCI Fixed 3 3 0x0/0x12 ata_intr, uhci_intr, ata_intr 19 0x22 1 PCI Fixed 1 1 0x0/0x13 uhci_intr 23 0x20 1 PCI Fixed 0 1 0x0/0x17 ehci_intr 24 0x45 5 PCI Fixed 0 1 0x1/0x0 adpu320_intr 48 0x60 6 PCI Fixed 2 1 0x2/0x0 e1000g_intr 72 0x43 5 MSI 3 1 - mpt_intr 73 0x44 5 MSI 3 1 - mpt_intr 160 0xa0 0 IPI ALL 0 - poke_cpu 192 0xc0 13 IPI ALL 1 - xc_serv 208 0xd0 14 IPI ALL 1 - kcpc_hw_overflow_intr 209 0xd1 14 IPI ALL 1 - cbe_fire 210 0xd3 14 IPI ALL 1 - cbe_fire 224 0xe0 15 IPI ALL 1 - xc_serv 225 0xe1 15 IPI ALL 1 - apic_error_intrWhere IPL is the interrupt priority, APIC is the local APIC,
Driver name is represented as <driver_name><instance#>
and Type shows where Fixed (legacy) or MSI interrupts are being used.
IPI interrupt type indicates xcalls.
The Share column shows if the interrupt is being shared and by how many.
On a uppc(7d) based system, the sample output is shown here
# echo ::interrupts | mdb -k > ::interrupts IRQ Vector IPL(lo/hi) Bus Share ISR(s) 0 0x20 14/14 - 1 cbe_fire 3 0x23 12/12 ISA 1 asyintr 4 0x24 12/12 ISA 1 asyintr 5 0x25 1/1 PCI 1 ehci_intr 6 0x26 5/5 ISA 1 fdc_intr 7 0x27 5/5 ISA 1 ecpp_isr 9 0x29 9/9 - 1 acpi_wrapper_isr 10 0x2a 1/9 PCI 3 e1000g_intr, hci1394_isr, uhci_intr 11 0x2b 1/1 PCI 1 uhci_intr 12 0x2c 1/9 PCI 2 uhci_intr, audiovia823x_intr 14 0x2e 5/5 PCI 1 ata_intr 15 0x2f 5/5 PCI 1 ata_intr # echo ::interrupts -d | mdb -k IRQ Vector IPL(lo/hi) Bus Share Driver Name(s) 0 0x20 14/14 - 1 cbe_fire 3 0x23 12/12 ISA 1 asy#1 4 0x24 12/12 ISA 1 asy#0 5 0x25 1/1 PCI 1 ehci#0 6 0x26 5/5 ISA 1 fdc#0 7 0x27 5/5 ISA 1 ecpp#0 9 0x29 9/9 - 1 acpi_wrapper_isr 10 0x2a 1/9 PCI 3 e1000g#0, hci1394#0, uhci#1 11 0x2b 1/1 PCI 1 uhci#0 12 0x2c 1/9 PCI 2 uhci#2, audiovia823x#0 14 0x2e 5/5 PCI 1 ata#0 15 0x2f 5/5 PCI 1 ata#1
On a Niagara based SunFire T2000, the output looks like this:
# echo ::interrupts | mdb -k Device Shared Type MSG # State INO Mondo Pil CPU e1000g#1 no MSI 1 enbl 0x19 0x799 6 26 e1000g#0 no MSI 0 enbl 0x18 0x798 6 18 px#0 no PCIe 27 enbl 0x3b 0x7bb 1 30 px#0 no PCIe 51 enbl 0x3a 0x7ba 14 31 px#0 no PCIe 49 enbl 0x39 0x7b9 14 0 px#0 no PCIe 48 enbl 0x38 0x7b8 9 1 e1000g#3 no MSI 2 enbl 0x1a 0x7da 6 28 e1000g#2 no MSI 1 enbl 0x19 0x7d9 6 27 su#0 no Fixed --- enbl 0x2 0x7c2 12 20 uata#0 no Fixed --- enbl 0x4 0x7c4 4 3 ohci#1 no Fixed --- enbl 0x3 0x7c3 9 22 ohci#0 no Fixed --- enbl 0x1 0x7c1 9 23 mpt#0 no MSI 0 enbl 0x18 0x7d8 4 19 px#1 no PCIe 27 enbl 0x3b 0x7fb 1 24 px#1 no PCIe 51 enbl 0x3a 0x7fa 14 25 px#1 no PCIe 49 enbl 0x39 0x7f9 14 26 px#1 no PCIe 48 enbl 0x38 0x7f8 9 27Here interrupt Type PCIe implies it is using PCI Express INTx.
On a SunFire V890, (which does not support MSI interrupts, the ouput looks like this:
# echo ::interrupts | mdb -k Device Shared Type MSG # State INO Mondo Pil CPU uata#0 no Fixed --- enbl 0x1c 0x21c 4 19 ge#0 no Fixed --- enbl 0x0 0x200 6 7 qlc#0 no Fixed --- enbl 0x4 0x204 4 18 su#1 no Fixed --- enbl 0x2d 0x26d 12 1 su#0 no Fixed --- enbl 0x2e 0x26e 12 0 pcf8584#2 yes Fixed --- enbl 0x28 0x268 4 2 pcf8584#3 yes Fixed --- enbl 0x28 0x268 4 2 eri#0 no Fixed --- enbl 0x1d 0x25d 6 17 ohci#0 no Fixed --- enbl 0x1f 0x25f 9 20 se#0 no Fixed --- enbl 0x22 0x262 12 21 todds1287#0 no Fixed --- enbl 0x24 0x264 15 22 hpc3130#3 no Fixed --- enbl 0x26 0x266 1 23 hpc3130#0 yes Fixed --- enbl 0x27 0x267 1 7 hpc3130#1 yes Fixed --- enbl 0x27 0x267 1 7 hpc3130#2 yes Fixed --- enbl 0x27 0x267 1 7 pcf8584#1 yes Fixed --- enbl 0x23 0x263 4 0 pcf8584#0 yes Fixed --- enbl 0x23 0x263 4 0
Technorati Tag: OpenSolaris Technorati Tag: Solaris Posted by anish ( Apr 14 2006, 08:29:15 AM PDT ) Permalink
Wednesday March 29, 2006 New Article on Advanced DDI interrupt interfaces
Advanced DDI Interrupt Handlers.
Thanks to John Stearns for making it possible.
Technorati Tag: OpenSolaris Technorati Tag: Solaris
Posted by anish ( Mar 29 2006, 01:54:26 PM PST ) Permalink
Tuesday February 28, 2006 Recent India Trip Feb 2006
I was in India and went to visit the holy Mata Vaishno Devi Shrine on Feb 18.
First I traveled by bus with a couple of friends to the town of Katra. We then
took an auto-rickshaw to Ban Ganga which forms the Holy Entrance for the trek.
The temple is located at 5200ft above sea level and is 13 kilometers (~8 miles) away from Ban Ganga.
One could walk along the serpentine paved path or walk up the steps all the way to the top.
We walked along the path and took the steps most of the way down.
Here is a View of Trikuta Hills taken about 50 miles away shows the serpentine trekking route
to the top of the hills ( see the white dotted lines). I have seen people run, crawl and roll up
the entire path. Our visit (or darshan) was successfully done within 8 hours.
On an unrelated topic, here is a picture of an Elephant near my house in Jammu Tawi.
Posted by anish ( Feb 28 2006, 02:53:47 PM PST ) Permalink
Monday January 30, 2006 See What's new for SX 1/06 for details on x86: PCI Express Support.
Technorati Tag: OpenSolaris Technorati Tag: Solaris
Posted by anish ( Jan 30 2006, 02:22:30 PM PST ) Permalink
Wednesday December 28, 2005 use MSIs (if the underlying hardware is capable) has increased. List includes:
- bge (Broadcom GbE)
- mpt (LSI Logic SCSI HBA)
- ehci/ohci/uhci (USB EHCI/OpenHCI/UHCI)
- e1000g (Intel GbE)
- tavor (InfiniBand HCA)
- px_pci (Sparc PCIe to PCI bridge driver)
- qlc (Qlogic FC HBA)
In addition, the following drivers use the new Advanced DDI Interrupt Framework natively.
- audio1575 (Sparc Audio driver )
- pcie_pci (x86 PCIe bridge driver)
- qcn (on Sparc servers)
- rge (Realtek GbE driver)
- nge (Nvidia GbE driver)
- SATA HBA drivers
- Support for PCIe extended configuration space
- New nexus drivers for PCIe (npe(7d)) and PCIe to PCI bridge (pcie_pci(7d))
- On PCIe systems npe(7d) replaces pci(7d)
- This can be viewed through prtconf(1m)
- Support for MSI interrupts and PCIe baseline error handling
- Modified IEEE-1275 properties (device_type shows 'pciex' )
- Any OpenSolaris drivers that look for device_type "pci" should
also look for "pciex"
- Support for PCIe hotplug (both ACPI based and Native PCIe).
- Provided via cfgadm(1m) CLI by modifying cfgadm_pci(1m)
- ATTN Button usage based PCIe peripheral auto-configuration
- Administrative model for PCIe hotplug is akin to PCI
Technorati Tag: OpenSolaris Technorati Tag: Solaris
Posted by anish ( Dec 28 2005, 06:47:50 PM PST ) Permalink Comments [0]
Tuesday December 27, 2005 "x86" Support for PCI Express (PCIe)
OpenSolaris b28 onwards provides support for the PCIe (PCIe) interconnect which is designed to interconnect peripheral devices to desktop, enterprise, mobile, communcation, and embedded applications. For details on PCIe serial I/O bus see PCI SIG.
With Opensolaris b28 release onwards the followin features are being provided
Posted by anish ( Dec 27 2005, 07:04:45 PM PST ) Permalink
Sunday December 25, 2005 See details at Jammu Tawi which has all the gory details ;-)
Posted by anish ( Dec 25 2005, 06:42:09 PM PST ) Permalink
Friday November 04, 2005 Solaris b26: changes to DDI interrupt interfaces
More legacy DDI interrupt interfaces are obsoleted
With Solaris b26 the following DDI Interrupt interfaces and data structures are being obsoleted:
- ddi_dev_nintrs(9f)
- ddi_get_iblock_cookie(9f)
- ddi_intr_hilevel(9f)
- ddi_add_intr(9f)
- ddi_remove_intr(9f)
- ddi_get_soft_iblock_cookie(9f)
- ddi_add_softintr(9f)
- ddi_remove_softintr(9f)
- ddi_trigger_softintr(9f)
- ddi_idevice_cookie(9s)
See modified Appendix B of the latest Solaris Express WDD that maps above obsoleted interfaces, data structures
to the new DDI Interrupt interfaces and data structures respectively.
We strongly urge the device driver developers to start using the new interrupt DDI Interfaces
Technorati Tag: OpenSolaris Technorati Tag: Solaris
Posted by anish ( Nov 04 2005, 11:47:28 AM PST ) Permalink Comments [1]
Monday July 11, 2005 Solaris Express 6/05 WDD released
Technorati Tag: OpenSolaris Technorati Tag: Solaris
Posted by anish ( Jul 11 2005, 02:14:43 PM PDT ) Permalink Comments [0]
| Archives | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Language | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Links | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Referrers | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Today's Page Hits: 23 |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
