T.R | Title | User | Personal Name | Date | Lines |
---|
9148.1 | | SMURF::DENHAM | Digital UNIX Kernel | Wed Mar 12 1997 21:17 | 12 |
| Sure. The kernel tried to shove some data onto the stack
of this thread to pass some scheduling information out
to the threads library in user space. This is to support
our 2-level threads scheuling scheme.
Well, there wasn't room on the stack, so we had to kill the
process. This is comparable to a similar message from the
signal subsystem when signal information can't be written
to the stack. The 2 mechanisms are quite similar.
Now, a first guess as to the problem is that your stack
is too small.
|
9148.2 | Still a problem | HYDRA::DAVIS | | Fri Mar 14 1997 11:46 | 25 |
| I have tried increasing the stack size but still get the error message.
Currently limit in csh gives
cputime unlimited
filesize unlimited
datasize 1048576 kbytes
stacksize 524287 kbytes
coredumpsize unlimited
memoryuse 1022560 kbytes
descriptors 4096 files
addressspace 1048576 kbytes
This problem comes from a software partner who is trying to qualify his
product on 4.0A. The image is built on 3.2?. The error message does not
show up on 3.2, only 4.0.
The message occurs when the product is shutting down and seems to have
no detrimental effect.
They can supply a reproducer.
Marvin Davis
[email protected]
DTN 297 6853
|
9148.3 | | SMURF::DENHAM | Digital UNIX Kernel | Fri Mar 14 1997 16:51 | 1 |
| Can only happen on V4.0 releases. Let's have the reproducer!
|
9148.4 | Did you increase the correct thread's stack size? | WTFN::SCALES | Despair is appropriate and inevitable. | Fri Mar 14 1997 16:53 | 13 |
| .2> I have tried increasing the stack size but still get the error message.
The problem is occuring on a thread stack; unless that thread just happens to be
the initial thread, you cannot increase it's stack size without changing the
application source code (or patching the image...), that is, the "limit" command
affects only the initial thread's stack.
.2> The error message does not show up on 3.2, only 4.0.
This is not surprising -- the "nxm" stuff is new in V4.0.
Webb
|
9148.5 | Partner's bug | HYDRA::DAVIS | | Thu Mar 20 1997 09:14 | 7 |
| The software partner reports that they have found the bug in their
code.
Thanks to all for your help
Marvin
|