-
Notifications
You must be signed in to change notification settings - Fork 179
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Clipping of vector slice results crashes smokeview #1563
Comments
Ill take a look
…On Tue, May 2, 2023, 4:57 AM obscureed ***@***.***> wrote:
If a vector slice is displayed in smokeview, and if clipping of results is
then activated, then smokeview immediately crashes with message "smokeview:
terminated by signal 11" or "Segmentation fault". This does not happen if
the clipping is of blockages only.
I've attached a test case that demonstrates this bug. Rename to .fds and
.ssf, then run "fds clip-vectors.fds" and "smokeview -script
clip-vectors.ssf clip-vectors.smv". On my machine, this produces two images
(vect-t0001.png with no clipping, vect-t0002.png with blockage clipping)
then crashes without producing the third image.
I see this on my current machine, running Ubuntu in WSL2 on Windows 11
Pro, for both the current smokeview (SMV-6.8.0-0-gd02fd6867-release) and
the one corresponding to FDS6.7.9 (SMV6.7.21-0-gccbc561da-release).
However, I am sure that I did not see this bug in FDS6.7.9 and SMV6.7.21 on
a previous machine (which I cannot now test) running Windows 10. So the
difference might be something connected to graphics engines, graphics
cards, etc.
On my current machine, smokeview also issues the "Segmentation fault"
message on quitting, even if no vector slice and no clipping is involved.
If I run smokeview on my current machine, but without GPU support, then
there is no crash or message. This is a diagnostic rather than a
workaround. Interestingly, if I run "glxinfo", it ends with "Segmentation
fault". I've attached the output from "glxinfo" and "glxinfo -B".
Thanks for your efforts,
Ed
glxinfo-B.txt
<https://github.com/firemodels/smv/files/11371464/glxinfo-B.txt>
glxinfo.txt <https://github.com/firemodels/smv/files/11371465/glxinfo.txt>
clip-vectors.ssf.txt
<https://github.com/firemodels/smv/files/11371131/clip-vectors.ssf.txt>
clip-vectors.fds.txt
<https://github.com/firemodels/smv/files/11371134/clip-vectors.fds.txt>
—
Reply to this email directly, view it on GitHub
<#1563>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AC6UCRVWYDFWFPY6NZY6UC3XEDEARANCNFSM6AAAAAAXSX5R5A>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
|
I not able to reproduce problem. I ran smokeview 6.8.0 test in windows 11 pro and Centos 7 ) with GPU both on and off. you ran smokeview 6.8.0 in a virutal linux command shell under windows 11 pro ? did you try running under windows 11 itself? |
I'll try running in a virtual ubuntu shell within windows 11 pro. perhaps
memory is an issue. do you know how much memory is allocated to your shell?
…On Tue, May 2, 2023 at 7:09 AM gforney ***@***.***> wrote:
I not able to reproduce problem. I ran smokeview 6.8.0 test in windows 11
pro and Centos 7 ) with GPU both on and off. you ran smokeview 6.8.0 in a
virutal linux command shell under windows 11 pro ? did you try running
under windows 11 itself?
—
Reply to this email directly, view it on GitHub
<#1563 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AC6UCRSHQJUHBEK5RIW54NDXEDTNHANCNFSM6AAAAAAXSX5R5A>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
--
Glenn Forney
|
I cannot reproduce the problem on a linux computer running Centos 7. |
I have not installed FDS as a Windows application -- I could try it as a test case, but it would impede my workflow to use it routinely. If it does or does not reproduce the bug, I'm not sure that it helps to clarify the situation -- we already know that the bug does not occur in all setups. I do not believe that memory is an issue. (Well, the error message is about segmentation fault, so memory is involved, but not memory sizes.) The machine has 32GB of RAM, and I do not think that Windows Subsystem for Linux is limited to any specific fraction of that. In particular, I do not think memory limits should be hit for the small test case that I posted. (I have processed much larger cases without any problems, so long as I don't try to clip vector slices.) I agree that the bug seems to be specific to some collection of OS and hardware -- it could be smokeview calling a GL command (for example) with syntax that is tolerated by some GL libraries/engines/renderers/whatever but not tolerated by my current setup. The problem could be smokeview or it could be the library/engine/etc. Is there any troubleshooting you can recommend, apart from trying to compile smokeview from source in debug mode? (I could try the debug compile, but I will not have time for it in the next month or two.) I use clipping fairly often, including to view vector slices, but mostly to clip the blockages that obscure the view. So, restricting the clip to blockages only is an acceptable workaround for me for now. |
can you clip other file types? ie regular slice files, boundary files, 3D
smoke files etc?
…On Tue, May 2, 2023 at 2:00 PM obscureed ***@***.***> wrote:
I have not installed FDS as a Windows application -- I could try it as a
test case, but it would impede my workflow to use it routinely. If it does
or does not reproduce the bug, I'm not sure that it helps to clarify the
situation -- we already know that the bug does not occur in all setups.
I do not believe that memory is an issue. (Well, the error message is
about segmentation fault, so memory is involved, but not memory sizes.) The
machine has 32GB of RAM, and I do not think that Windows Subsystem for
Linux is limited to any specific fraction of that. In particular, I do not
think memory limits should be hit for the small test case that I posted. (I
have processed much larger cases without any problems, so long as I don't
try to clip vector slices.)
I agree that the bug seems to be specific to some collection of OS and
hardware -- it could be smokeview calling a GL command (for example) with
syntax that is tolerated by some GL libraries/engines/renderers/whatever
but not tolerated by my current setup. The problem could be smokeview or it
could be the library/engine/etc. Is there any troubleshooting you can
recommend, apart from trying to compile smokeview from source in debug
mode? (I could try the debug compile, but I will not have time for it in
the next month or two.)
I use clipping fairly often, including to view vector slices, but mostly
to clip the blockages that obscure the view. So, restricting the clip to
blockages only is an acceptable workaround for me for now.
—
Reply to this email directly, view it on GitHub
<#1563 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AC6UCRRTQBCF6ALBXLU4OTTXEFDUZANCNFSM6AAAAAAXSX5R5A>
.
You are receiving this because you were assigned.Message ID:
***@***.***>
--
Glenn Forney
|
Yes to regular slice files. Yes to 3D smoke -- but that requires "Use GPU" to be deselected, and vector slices work then as well. (With "Use GPU" selected, the smoke disappears, as expected, and there is no crash.) I haven't tried boundary files yet. |
Here is another 'Segmentation fault' crash: I've added this to this Issue, but I have a feeling it could be a different bug entirely, because I can reproduce this on an Ubuntu machine. The keyboard shortcut '&' is here intended to toggle between HVAC metro view modes. I think a part of the problem here is that there are no HVAC elements in the case. I have attached a modified test case with an HVAC VENT, and this does not crash with &. |
does it crash with the smokeview where I reverted '&' back to toggling
antialiasing ?
…On Fri, May 5, 2023 at 2:12 PM obscureed ***@***.***> wrote:
Here is another 'Segmentation fault' crash:
• run the test case above (clip-vectors.fds);
• start smokeview interactively;
• press '&'.
I've added this to this Issue, but I have a feeling it could be a
different bug entirely, because I can reproduce this on an Ubuntu machine.
The keyboard shortcut '&' is here intended to toggle between HVAC metro
view modes. I think a part of the problem here is that there are no HVAC
elements in the case. I have attached a modified test case with an HVAC
VENT, and this does not crash with &.
clip-vectors_with-hvac.fds.txt
<https://github.com/firemodels/smv/files/11409014/clip-vectors_with-hvac.fds.txt>
—
Reply to this email directly, view it on GitHub
<#1563 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AC6UCRVP6OFSULMLUDK2LS3XEU7HPANCNFSM6AAAAAAXSX5R5A>
.
You are receiving this because you were assigned.Message ID:
***@***.***>
--
Glenn Forney
|
should be fixed in smokeview posted here |
In that patched version of smokeview (SMV-6.8.0-110-gab3231e6d-master): |
unfortunately I am not able to reproduce the problem on the windows, mac
and linux computers we use. I'll see if I can setup a linux virtual
machine on my windows computer and try it there
…On Thu, May 11, 2023 at 11:24 AM obscureed ***@***.***> wrote:
In that patched version of smokeview (SMV-6.8.0-110-gab3231e6d-master):
• The '&' keyboard shortcut is back to toggling antialiasflag, and does
not cause a crash.
• Clipping a vector slice still crashes smokeview, at least on my specific
machine.
—
Reply to this email directly, view it on GitHub
<#1563 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AC6UCRQIYQYCV7IS6RXXQCLXFUABXANCNFSM6AAAAAAXSX5R5A>
.
You are receiving this because you were assigned.Message ID:
***@***.***>
--
Glenn Forney
|
Just for information, I still get this crash with SMV 6.9.0. I expect you still can't reproduce it, so there's no prospect of it being fixed -- and I suspect it's a wider issue in some libraries or whatever. My workaround is still to not clip vector slices. |
Ill look some more
…On Fri, Apr 12, 2024, 12:53 PM obscureed ***@***.***> wrote:
Just for information, I still get this crash with SMV 6.9.0. I expect you
still can't reproduce it, so there's no prospect of it being fixed -- and I
suspect it's a wider issue in some libraries or whatever. My workaround is
still to not clip vector slices.
—
Reply to this email directly, view it on GitHub
<#1563 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AC6UCRQQP4J3GIQTR5KWMQLY5AGRNAVCNFSM6AAAAAAXSX5R5CVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDANJSGEZDEMRVHE>
.
You are receiving this because you were assigned.Message ID:
***@***.***>
|
If a vector slice is displayed in smokeview, and if clipping of results is then activated, then smokeview immediately crashes with message "smokeview: terminated by signal 11" or "Segmentation fault". This does not happen if the clipping is of blockages only.
I've attached a test case that demonstrates this bug. Rename to .fds and .ssf, then run "fds clip-vectors.fds" and "smokeview -script clip-vectors.ssf clip-vectors.smv". On my machine, this produces two images (vect-t0001.png with no clipping, vect-t0002.png with blockage clipping) then crashes without producing the third image.
I see this on my current machine, running Ubuntu in WSL2 on Windows 11 Pro, for both the current smokeview (SMV-6.8.0-0-gd02fd6867-release) and the one corresponding to FDS6.7.9 (SMV6.7.21-0-gccbc561da-release). However, I am sure that I did not see this bug in FDS6.7.9 and SMV6.7.21 on a previous machine (which I cannot now test) running Windows 10. So the difference might be something connected to graphics engines, graphics cards, etc.
On my current machine, smokeview also issues the "Segmentation fault" message on quitting, even if no vector slice and no clipping is involved.
If I run smokeview on my current machine, but without GPU support, then there is no crash or message. This is a diagnostic rather than a workaround. Interestingly, if I run "glxinfo", it ends with "Segmentation fault". I've attached the output from "glxinfo" and "glxinfo -B".
Thanks for your efforts,
Ed
glxinfo-B.txt
glxinfo.txt
clip-vectors.ssf.txt
clip-vectors.fds.txt
The text was updated successfully, but these errors were encountered: