Hi all,
Considering that I have a target host that hasn’t any serial port.
Can I use a PCI adapter that creates new serial ports in this host
instead of buying a 1394 Card for Debugging?
Thanks in advance,
Fernando Roberto da Silva
Hi all,
Considering that I have a target host that hasn’t any serial port.
Can I use a PCI adapter that creates new serial ports in this host
instead of buying a 1394 Card for Debugging?
Thanks in advance,
Fernando Roberto da Silva
The kernel debugging support has hardcoded support for UARTs that look like a 16550, and expects to be able to communicate with them at the standard I/O addresses, 0x3f8 (for COM1) and 0x2f8 (for COM2). (Not sure about COM3/COM4/etc.) The KD support in the kernel does not use a loadable driver, so if you want to use a serial UART, this is your only option. If your PCI board can be configured to work this way, and be ready to do so at OS boot time, then you can probably get it to work.
However, that sounds a lot less fun than just buying a 1394 board. Considering the “good” TI boards cost US$29, that doesn’t seem too onerous, even including international shipping, and there’s probably somewhere nearer you that has them available.
– arlie
From: xxxxx@lists.osr.com [mailto:xxxxx@lists.osr.com] On Behalf Of Fernando Roberto
Sent: Wednesday, October 18, 2006 10:49 AM
To: Kernel Debugging Interest List
Subject: [windbg] Windbg through PCI serial ports
Hi all,
Considering that I have a target host that hasn’t any serial port.
Can I use a PCI adapter that creates new serial ports in this host instead of buying a 1394 Card for Debugging?
Thanks in advance,
Fernando Roberto da Silva
You are currently subscribed to windbg as: unknown lmsubst tag argument: ‘’
To unsubscribe send a blank email to xxxxx@lists.osr.com
This kind of thing (blades with no legacy support) is a perennial
problem for us (our product is typically deployed in datacenters and
those admins love blades. Typically these things have one PCI slot at
best (some have none). Our problem is made worse by the fact that we
usually need something in that single slot (a fibre-channel HBA).
So debugging this class of hardware is difficult as best. Are there any
plans to extend debugging to other types of connection (say, Ethernet
MAC, which you can pretty much guarantee everything out there has)? This
kind of hardware is becoming more popular, not less popular, so this
would seem to be a good thing to do for the future…
…dave
(and before you say “repro the problem on some other hardware”, we have
(more than once) run into issues that are specific to blade hardware,
and which don’t repro on other hardware. So this isn’t an academic
exercise, it’s a very real-world problem.)
-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of Arlie Davis
Sent: Wednesday, October 18, 2006 11:24 AM
To: Kernel Debugging Interest List
Subject: RE:[windbg] Windbg through PCI serial ports
The kernel debugging support has hardcoded support for UARTs that look
like a 16550, and expects to be able to communicate with them at the
standard I/O addresses, 0x3f8 (for COM1) and 0x2f8 (for COM2). (Not
sure about COM3/COM4/etc.) The KD support in the kernel does not use a
loadable driver, so if you want to use a serial UART, this is your only
option. If your PCI board can be configured to work this way, and be
ready to do so at OS boot time, then you can probably get it to work.
However, that sounds a lot less fun than just buying a 1394 board.
Considering the “good” TI boards cost US$29, that doesn’t seem too
onerous, even including international shipping, and there’s probably
somewhere nearer you that has them available.
– arlie
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of Fernando Roberto
Sent: Wednesday, October 18, 2006 10:49 AM
To: Kernel Debugging Interest List
Subject: [windbg] Windbg through PCI serial ports
Hi all,
Considering that I have a target host that hasn’t any serial port.
Can I use a PCI adapter that creates new serial ports in this host
instead of buying a 1394 Card for Debugging?
Thanks in advance,
Fernando Roberto da Silva
You are currently subscribed to windbg as: unknown lmsubst tag argument:
‘’
To unsubscribe send a blank email to xxxxx@lists.osr.com
You are currently subscribed to windbg as: unknown lmsubst tag argument:
‘’
To unsubscribe send a blank email to xxxxx@lists.osr.com
If you run Vista or later, USB is available. Otherwise, I suspect serial
and 1394a are it. I am surprised that blades have eliminated the serial
ports. I thought most of them used those for UPS monitoring, but maybe they
have moved to USB. If you can, check the motherboard and see if jumpers for
serial or 1394 might be available.
“Dave Beaver” wrote in message news:xxxxx@windbg…
This kind of thing (blades with no legacy support) is a perennial
problem for us (our product is typically deployed in datacenters and
those admins love blades. Typically these things have one PCI slot at
best (some have none). Our problem is made worse by the fact that we
usually need something in that single slot (a fibre-channel HBA).
So debugging this class of hardware is difficult as best. Are there any
plans to extend debugging to other types of connection (say, Ethernet
MAC, which you can pretty much guarantee everything out there has)? This
kind of hardware is becoming more popular, not less popular, so this
would seem to be a good thing to do for the future…
…dave
(and before you say “repro the problem on some other hardware”, we have
(more than once) run into issues that are specific to blade hardware,
and which don’t repro on other hardware. So this isn’t an academic
exercise, it’s a very real-world problem.)
-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of Arlie Davis
Sent: Wednesday, October 18, 2006 11:24 AM
To: Kernel Debugging Interest List
Subject: RE:[windbg] Windbg through PCI serial ports
The kernel debugging support has hardcoded support for UARTs that look
like a 16550, and expects to be able to communicate with them at the
standard I/O addresses, 0x3f8 (for COM1) and 0x2f8 (for COM2). (Not
sure about COM3/COM4/etc.) The KD support in the kernel does not use a
loadable driver, so if you want to use a serial UART, this is your only
option. If your PCI board can be configured to work this way, and be
ready to do so at OS boot time, then you can probably get it to work.
However, that sounds a lot less fun than just buying a 1394 board.
Considering the “good” TI boards cost US$29, that doesn’t seem too
onerous, even including international shipping, and there’s probably
somewhere nearer you that has them available.
– arlie
________________________________
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of Fernando Roberto
Sent: Wednesday, October 18, 2006 10:49 AM
To: Kernel Debugging Interest List
Subject: [windbg] Windbg through PCI serial ports
Hi all,
Considering that I have a target host that hasn’t any serial port.
Can I use a PCI adapter that creates new serial ports in this host
instead of buying a 1394 Card for Debugging?
Thanks in advance,
Fernando Roberto da Silva
http://www.driverentry.com.br
—
You are currently subscribed to windbg as: unknown lmsubst tag argument:
‘’
To unsubscribe send a blank email to xxxxx@lists.osr.com
—
You are currently subscribed to windbg as: unknown lmsubst tag argument:
‘’
To unsubscribe send a blank email to xxxxx@lists.osr.com
Microsoft is looking at other kinds of transports for kernel debugging,
such as Ethernet, but there’s no timetable for anything. Right now your
only options are serial, 1394 and USB 2.0.
-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of Dave Beaver
Sent: Wednesday, October 18, 2006 10:35 PM
To: Kernel Debugging Interest List
Subject: RE: [windbg] Windbg through PCI serial ports
This kind of thing (blades with no legacy support) is a perennial
problem for us (our product is typically deployed in datacenters and
those admins love blades. Typically these things have one PCI slot at
best (some have none). Our problem is made worse by the fact that we
usually need something in that single slot (a fibre-channel HBA).
So debugging this class of hardware is difficult as best. Are there any
plans to extend debugging to other types of connection (say, Ethernet
MAC, which you can pretty much guarantee everything out there has)? This
kind of hardware is becoming more popular, not less popular, so this
would seem to be a good thing to do for the future…
…dave
(and before you say “repro the problem on some other hardware”, we have
(more than once) run into issues that are specific to blade hardware,
and which don’t repro on other hardware. So this isn’t an academic
exercise, it’s a very real-world problem.)
-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of Arlie Davis
Sent: Wednesday, October 18, 2006 11:24 AM
To: Kernel Debugging Interest List
Subject: RE:[windbg] Windbg through PCI serial ports
The kernel debugging support has hardcoded support for UARTs that look
like a 16550, and expects to be able to communicate with them at the
standard I/O addresses, 0x3f8 (for COM1) and 0x2f8 (for COM2). (Not
sure about COM3/COM4/etc.) The KD support in the kernel does not use a
loadable driver, so if you want to use a serial UART, this is your only
option. If your PCI board can be configured to work this way, and be
ready to do so at OS boot time, then you can probably get it to work.
However, that sounds a lot less fun than just buying a 1394 board.
Considering the “good” TI boards cost US$29, that doesn’t seem too
onerous, even including international shipping, and there’s probably
somewhere nearer you that has them available.
– arlie
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of Fernando Roberto
Sent: Wednesday, October 18, 2006 10:49 AM
To: Kernel Debugging Interest List
Subject: [windbg] Windbg through PCI serial ports
Hi all,
Considering that I have a target host that hasn’t any serial port.
Can I use a PCI adapter that creates new serial ports in this host
instead of buying a 1394 Card for Debugging?
Thanks in advance,
Fernando Roberto da Silva
You are currently subscribed to windbg as: unknown lmsubst tag argument:
‘’
To unsubscribe send a blank email to xxxxx@lists.osr.com
You are currently subscribed to windbg as: unknown lmsubst tag argument:
‘’
To unsubscribe send a blank email to xxxxx@lists.osr.com
You are currently subscribed to windbg as: unknown lmsubst tag argument:
‘’
To unsubscribe send a blank email to xxxxx@lists.osr.com
Dave,
In my experience (working at an OEM that builds such systems) there is
always a uart on the system somewhere to enable kernel debugging. Keep
in mind that its not only Windows driver developers that need this
hardware support, it is also used by BIOS developers, other operating
systems, etc… Usually in the case of IBM/Intel blades you need an
extender cable that runs from a small un-noticed socket on the baseboard
and squeezes through a crack in the chassis to the front panel…
-Zach
-----Original Message-----
From: xxxxx@lists.osr.com [mailto:bounce-266954-
xxxxx@lists.osr.com] On Behalf Of Dave Beaver
Sent: Wednesday, October 18, 2006 10:35 PM
To: Kernel Debugging Interest List
Subject: RE: [windbg] Windbg through PCI serial portsThis kind of thing (blades with no legacy support) is a perennial
problem for us (our product is typically deployed in datacenters and
those admins love blades. Typically these things have one PCI slot at
best (some have none). Our problem is made worse by the fact that we
usually need something in that single slot (a fibre-channel HBA).