Re: [mapserver-users] AGG vs GD rendering shows no difference
I've managed to run mapserv through gdb, and my suspicions were
confirmed - it dies in the AGG part:
...
Reading symbols from shared object read from target memory...done.
Loaded system supplied DSO at 0xa3e000
[Thread debugging using libthread_db enabled]
[New Thread -1208244032 (LWP 18090)]
*Program received signal SIGSEGV, Segmentation fault.*
[Switching to Thread -1208244032 (LWP 18090)]
*0x00402b8c in agg::vertex_sequenceagg::vertex_dist, 6u::add () from
/usr/lib/libagg.so.2*
(gdb) where
#0 0x00402b8c in agg::vertex_sequenceagg::vertex_dist, 6u::add ()
from /usr/lib/libagg.so.2
#1 0x00402c2f in agg::vertex_sequenceagg::vertex_dist,
6u::modify_last () from /usr/lib/libagg.so.2
#2 0x004050e5 in agg::vcgen_stroke::add_vertex () from /usr/lib/libagg.so.2
#3 0x080d2f4c in agg::conv_adaptor_vcgenline_adaptor,
agg::vcgen_stroke, agg::null_markers::vertex (this=0xbfdcdbf4,
x=0xbfdcdc98, y=0xbfdcdca0)
at
/home/adrianp/mapserver/mapserver-5.4.1/../agg-2.5/include/agg_conv_adaptor_vcgen.h:115
#4 0x080e4206 in AGGMapserverRenderer::renderPolylineline_adaptor
(this=0xa079968, p...@0xa081b58, col...@0xbfdcdf48, width=1,
dashstylelength=0, dashstyle=0x9c27980,
lc=agg::round_cap, lj=agg::round_join) at
/home/adrianp/mapserver/mapserver-5.4.1/../agg-2.5/include/agg_rasterizer_scanline_aa.h:206
#5 0x080ccc46 in msDrawLineSymbolAGG (symbolset=0x9c2556c,
image=0x9c4bb78, p=0xbfdce298, style=0x9c3ba18, scalefactor=1) at
mapagg.cpp:1672
#6 0x080b331d in msDrawLineSymbol (symbolset=0x9c2556c,
image=0x9c4bb78, p=0xbfdce298, style=0x9c3ba18, scalefactor=1) at
mapdraw.c:2117
#7 0x080b1d4d in msDrawShape (map=0x9c25548, layer=0x9c3ae60,
shape=0xbfdce298, image=0x9c4bb78, style=0, querymapMode=1) at
mapdraw.c:1782
#8 0x080ae1eb in msDrawVectorLayer (map=0x9c25548, layer=0x9c3ae60,
image=0x9c4bb78) at mapdraw.c:938
#9 0x080ad91d in msDrawLayer (map=0x9c25548, layer=0x9c3ae60,
image=0x9c4bb78) at mapdraw.c:743
#10 0x080acb54 in msDrawMap (map=0x9c25548, querymap=0) at mapdraw.c:441
#11 0x080584b3 in main (argc=2, argv=0xbfdcea84) at mapserv.c:1417
I'll have to check to see if there are known bugs regarding AGG...
Adrian Popa wrote:
Thank you for your suggestion Andreas,
I have run shp2img and unfortunately:
[r...@terra tmp]# shp2img -o /tmp/out.png -m
/var/www/html/msmap/dynamicmaps/fibUaBrx.map -e 20.258
43.16 29.703 49.2016 -s 1250 800 -l GranitaJudete
Judete Urban Rural RuralSate buildings fibra fiber-common fiber-STM64
fiber-STM16 fiber-STM4 fiber-STM1 fiber-problems-main
fiber-problems-related fibraProblems
*Segmentation fault*
Running the command through strace, it ends up like this:
read(4, h\376\237\224O\320;@p\225\4\25\'x...@\264d\270^\320;@\26...,
45056) = 45056
read(4, \303\362oU\36\206;@by\237\354...@\34\271\214\333\37\206...,
4096) = 4096
_llseek(9, 12288, [12288], SEEK_SET)= 0
read(9, ..., 4096) = 148
_llseek(4, 1458176, [1458176], SEEK_SET) = 0
read(4, \25\327\377\332\27\...@\30\256\23v9|;@\233M%\235\372\22...,
24576) = 24576
read(4, a\1$\203\...@.\245\211px
;@\314\353\330\355\242\...@\367..., 4096) = 1784
close(8)= 0
munmap(0xb7f36000, 4096)= 0
close(4)= 0
munmap(0xb7f38000, 4096)= 0
close(9)= 0
munmap(0xb7f35000, 4096)= 0
gettimeofday({1245157506, 462482}, NULL) = 0
gettimeofday({1245157506, 462600}, NULL) = 0
open(/etc/localtime, O_RDONLY)= 4
fstat64(4, {st_mode=S_IFREG|0644, st_size=798, ...}) = 0
fstat64(4, {st_mode=S_IFREG|0644, st_size=798, ...}) = 0
mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1,
0) = 0xb7f38000
read(4, TZif\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\7\0\0\0\7\0...,
4096) = 798
close(4)= 0
munmap(0xb7f38000, 4096)= 0
gettimeofday({1245157506, 463468}, NULL) = 0
gettimeofday({1245157506, 463535}, NULL) = 0
gettimeofday({1245157506, 463599}, NULL) = 0
gettimeofday({1245157506, 463664}, NULL) = 0
gettimeofday({1245157506, 463734}, NULL) = 0
gettimeofday({1245157506, 463802}, NULL) = 0
gettimeofday({1245157506, 463866}, NULL) = 0
gettimeofday({1245157506, 463934}, NULL) = 0
gettimeofday({1245157506, 464028}, NULL) = 0
*open(/var/www/html/msmap/dynamicmaps/../data/Judete.shp,
O_RDONLY|O_LARGEFILE) = 4*
open(/var/www/html/msmap/dynamicmaps/../data/Judete.shx,
O_RDONLY|O_LARGEFILE) = 8
fstat64(4, {st_mode=S_IFREG|0555, st_size=1484536, ...}) = 0
mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1,
0) = 0xb7f38000
*read(4*, \0\0\'\n\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\vS|...,
4096) = 4096
fstat64(8, {st_mode=S_IFREG|0555, st_size=436, ...}) = 0
mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1,
0) = 0xb7f36000
read(8, \0\0\'\n\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0...,
4096) = 436
Re: [mapserver-users] AGG vs GD rendering shows no difference
Hello everyone! After a few hours of digging, I managed to solve the problem. I checked mapserver's dependencies and it was linked against agg 2.3: [r...@terra tmp]# ldd /var/www/cgi-bin/mapserv | grep agg libagg.so.2 = /usr/lib/libagg.so.2 (0x003f5000) lrwxrwxrwx 1 root root 15 Jun 15 10:31 /usr/lib/libagg.so - libagg.so.2.0.1 lrwxrwxrwx 1 root root 15 Jun 15 10:31 /usr/lib/libagg.so.2 - libagg.so.2.0.1 -rwxr-xr-x 1 root root 228524 Feb 17 2006 /usr/lib/libagg.so.2.0.1 However, when I compiled mapserver I pointed the AGG code (--with-agg) to version 2.5. So mapserver had been compiled against agg 2.5 but was running with agg 2.3. This caused the segfault when using agg. The solution is to compile AGG2.5 and to install it in the system. Unfortunately, the AGG documentation is scarce and doesn't clearly say how to compile the code, and even if I'm a linux seasoned user, I failed to see the autogen.sh script. Here's how I compiled AGG: *sh autogen.sh ./configure --disable-platform --disable-examples --prefix=/usr make make install * And now, I have also the newer libs in my /usr/lib/ directory: [r...@terra mapserver]# ls -l /usr/lib/libagg* -rw-r--r-- 1 root root 1313570 Jun 17 17:06 /usr/lib/libagg.a -rw-r--r-- 1 root root 395874 Jun 17 17:06 /usr/lib/libaggfontfreetype.a -rwxr-xr-x 1 root root 897 Jun 17 17:06 /usr/lib/libaggfontfreetype.la lrwxrwxrwx 1 root root 27 Jun 17 17:06 /usr/lib/libaggfontfreetype.so - libaggfontfreetype.so.2.0.4 lrwxrwxrwx 1 root root 27 Jun 17 17:06 /usr/lib/libaggfontfreetype.so.2 - libaggfontfreetype.so.2.0.4 -rwxr-xr-x 1 root root 283338 Jun 17 17:06 /usr/lib/libaggfontfreetype.so.2.0.4 -rwxr-xr-x 1 root root 785 Jun 17 17:06 /usr/lib/libagg.la -rw-r--r-- 1 root root 53282 Feb 17 2006 /usr/lib/libaggplatformX11.a lrwxrwxrwx 1 root root 26 Jun 15 10:31 /usr/lib/libaggplatformX11.so - libaggplatformX11.so.2.0.1 lrwxrwxrwx 1 root root 26 Jun 15 10:31 /usr/lib/libaggplatformX11.so.2 - libaggplatformX11.so.2.0.1 -rwxr-xr-x 1 root root 43468 Feb 17 2006 /usr/lib/libaggplatformX11.so.2.0.1 lrwxrwxrwx 1 root root 15 Jun 17 17:06 /usr/lib/libagg.so - libagg.so.2.0.4 lrwxrwxrwx 1 root root 15 Jun 17 17:06 /usr/lib/libagg.so.2 - libagg.so.2.0.4 -rwxr-xr-x 1 root root 228524 Feb 17 2006 /usr/lib/libagg.so.2.0.1 -rwxr-xr-x 1 root root 774469 Jun 17 17:06 /usr/lib/libagg.so.2.0.4 And mapserver renders my maps very well (see attached photos) Thank you all for your help! Regards, Adrian inline: agg.pnginline: png.png___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users
Re: [mapserver-users] AGG vs GD rendering shows no difference
Adrian Popa wrote: The original mapfile had rendering set up like this: IMAGETYPE PNG while the new mapfile has rendering set up like this: IMAGETYPE PNG OUTPUTFORMAT NAME 'AGG' DRIVER AGG/PNG IMAGEMODE RGB END Adrian, you'll most likely need to write IMAGETYPE 'AGG' instead of IMAGETYPE PNG to make sure your new outputformat is used. Best regards, -- Andreas Albarello Analysis SW Development Territorium Online srl/GmbH Via Buozzi Str. 12 I 39100 Bolzano/Bozen Phone: +39 0471 068611 Fax:+39 0471 068619 email: andreas.albare...@territoriumonline.com web: http://www.territoriumonline.com ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users
Re: [mapserver-users] AGG vs GD rendering shows no difference
Thank you for your suggestion. I have corrected the mapfile ( added IMAGETYPE 'AGG'), however, when I try to load the map, I get a 500 web server error: [Tue Jun 16 13:43:25 2009] [error] [client 80.97.196.77] Premature end of script headers: mapserv, referer: http://terra/msmap/testmap.pl It seems to me that mapserv dies and doesn't output any headers back. How could I find out the error message? I already have a log statement in my WEB section: LOG /var/www/html/map.log (the file is writable by apache) Unfortunately I don't get anything in this log. My mapserver supports: [r...@terra html]# /var/www/cgi-bin/mapserv -v MapServer version 5.4.1 OUTPUT=GIF OUTPUT=PNG OUTPUT=JPEG OUTPUT=WBMP OUTPUT=SVG SUPPORTS=PROJ SUPPORTS=AGG SUPPORTS=FREETYPE SUPPORTS=ICONV SUPPORTS=WMS_SERVER SUPPORTS=WMS_CLIENT SUPPORTS=WFS_SERVER SUPPORTS=WFS_CLIENT SUPPORTS=WCS_SERVER SUPPORTS=SOS_SERVER SUPPORTS=THREADS SUPPORTS=GEOS INPUT=EPPL7 INPUT=POSTGIS INPUT=OGR INPUT=GDAL INPUT=MYGIS INPUT=SHAPEFILE Any ideas how to debug this? Thanks Andreas Albarello wrote: Adrian Popa wrote: The original mapfile had rendering set up like this: IMAGETYPE PNG while the new mapfile has rendering set up like this: IMAGETYPE PNG OUTPUTFORMAT NAME 'AGG' DRIVER AGG/PNG IMAGEMODE RGB END Adrian, you'll most likely need to write IMAGETYPE 'AGG' instead of IMAGETYPE PNG to make sure your new outputformat is used. Best regards, ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users
Re: [mapserver-users] AGG vs GD rendering shows no difference
On Tue, 16 Jun 2009 20:58:22 +1000, Adrian Popa adrian_gh.p...@romtelecom.ro wrote: Thank you for your suggestion. I have corrected the mapfile ( added IMAGETYPE 'AGG'), however, when I try to load the map, I get a 500 web server error: [Tue Jun 16 13:43:25 2009] [error] [client 80.97.196.77] Premature end of script headers: mapserv, referer: http://terra/msmap/testmap.pl It seems to me that mapserv dies and doesn't output any headers back. How could I find out the error message? The first place to check is the web server log. HTTP errors (and 500 is one) are usually visible there. I'm guessing that mapserver has a problem parsing the map file and you can usually discover the line it is having a problem with in the web server's log. The second level of logging is: CONFIG MS_ERRORFILE path_to_mapserver.log which provides more detail, but you usually have to configure DEBUG ON (or DEBUG 1-5) as well for this to be useful. Regards, David ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users
Re: [mapserver-users] AGG vs GD rendering shows no difference
On 16 Jun 2009, at 14:25, Adrian Popa wrote: Hello David, I got the 500 errors from the http error log file. Unfortunately no additional information is provided. I added the MS_ERRORFILE and DEBUG statements to my map file and although the errorfile is created, I don't get any other information (neither when the map loads perfectly, nor when it dies with premature end of script headers. I read RFC 28 (http://mapserver.org/development/rfc/ms-rfc-28.html) and they specified that the server should be compiled with - DENABLE_STDERR_DEBUG, so I recompiled mapserver: CFLAGS=-DENABLE_STDERR_DEBUG ./configure --with-freetype --with-png --with-agg=../agg-2.5 --with-proj --with-ogr --with-gdal --with-xml2 --with-wfs --with-wcs --with-wmsclient --with-wfsclient --with- postgis --with-threads --with-sos --with-mygis --with-geos --with-tiff I used the new binary and unfortunately nothing changed :( My (relevant part of the) mapfile looks like this: MAP CONFIG MS_ERRORFILE /tmp/mapserver.log CONFIG MS_DEBUGLEVEL 5 DEBUG 5... END What else can I try? Try to bypass the webserver/CGI stuff and launch shp2img directly from the shell, that should hopefully yield some better output: shp2img -o test.png -m mapfile.map If you haven't already done so in your mapfile, you might need to add an image size (-s 500 500) and/or an extent (-e minx miny maxx maxy) to the above command line options. Best regards, -- Andreas Albarello Analysis SW Development Territorium Online srl/GmbH Via Buozzi/Buozzistraße 12 - I 39100 Bolzano/Bozen email: andreas.albare...@territoriumonline.com web: www.territoriumonline.com ___ mapserver-users mailing list mapserver-users@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/mapserver-users
Re: [mapserver-users] AGG vs GD rendering shows no difference
Thank you for your suggestion Andreas,
I have run shp2img and unfortunately:
[r...@terra tmp]# shp2img -o /tmp/out.png -m
/var/www/html/msmap/dynamicmaps/fibUaBrx.map -e 20.258
43.16 29.703 49.2016 -s 1250 800 -l GranitaJudete Judete
Urban Rural RuralSate buildings fibra fiber-common fiber-STM64
fiber-STM16 fiber-STM4 fiber-STM1 fiber-problems-main
fiber-problems-related fibraProblems
*Segmentation fault*
Running the command through strace, it ends up like this:
read(4, h\376\237\224O\320;@p\225\4\25\'x...@\264d\270^\320;@\26...,
45056) = 45056
read(4, \303\362oU\36\206;@by\237\354...@\34\271\214\333\37\206...,
4096) = 4096
_llseek(9, 12288, [12288], SEEK_SET)= 0
read(9, ..., 4096) = 148
_llseek(4, 1458176, [1458176], SEEK_SET) = 0
read(4, \25\327\377\332\27\...@\30\256\23v9|;@\233M%\235\372\22...,
24576) = 24576
read(4, a\1$\203\...@.\245\211px ;@\314\353\330\355\242\...@\367...,
4096) = 1784
close(8)= 0
munmap(0xb7f36000, 4096)= 0
close(4)= 0
munmap(0xb7f38000, 4096)= 0
close(9)= 0
munmap(0xb7f35000, 4096)= 0
gettimeofday({1245157506, 462482}, NULL) = 0
gettimeofday({1245157506, 462600}, NULL) = 0
open(/etc/localtime, O_RDONLY)= 4
fstat64(4, {st_mode=S_IFREG|0644, st_size=798, ...}) = 0
fstat64(4, {st_mode=S_IFREG|0644, st_size=798, ...}) = 0
mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1,
0) = 0xb7f38000
read(4, TZif\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\7\0\0\0\7\0...,
4096) = 798
close(4)= 0
munmap(0xb7f38000, 4096)= 0
gettimeofday({1245157506, 463468}, NULL) = 0
gettimeofday({1245157506, 463535}, NULL) = 0
gettimeofday({1245157506, 463599}, NULL) = 0
gettimeofday({1245157506, 463664}, NULL) = 0
gettimeofday({1245157506, 463734}, NULL) = 0
gettimeofday({1245157506, 463802}, NULL) = 0
gettimeofday({1245157506, 463866}, NULL) = 0
gettimeofday({1245157506, 463934}, NULL) = 0
gettimeofday({1245157506, 464028}, NULL) = 0
*open(/var/www/html/msmap/dynamicmaps/../data/Judete.shp,
O_RDONLY|O_LARGEFILE) = 4*
open(/var/www/html/msmap/dynamicmaps/../data/Judete.shx,
O_RDONLY|O_LARGEFILE) = 8
fstat64(4, {st_mode=S_IFREG|0555, st_size=1484536, ...}) = 0
mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1,
0) = 0xb7f38000
*read(4*, \0\0\'\n\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\vS|...,
4096) = 4096
fstat64(8, {st_mode=S_IFREG|0555, st_size=436, ...}) = 0
mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1,
0) = 0xb7f36000
read(8, \0\0\'\n\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0...,
4096) = 436
open(/var/www/html/msmap/dynamicmaps/../data/Judete.dbf,
O_RDONLY|O_LARGEFILE) = 9
fstat64(9, {st_mode=S_IFREG|0555, st_size=12436, ...}) = 0
mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1,
0) = 0xb7f35000
read(9, \3k\6\24*\0\0\0\201\0%\1\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0...,
4096) = 4096
_llseek(9, 0, [0], SEEK_SET)= 0
read(9, \3k\6\24*\0\0\0\201\0%\1\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0...,
4096) = 4096
_llseek(8, 0, [0], SEEK_SET)= 0
read(8, \0\0\'\n\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0...,
4096) = 436
read(8, , 4096) = 0
_llseek(4, 0, [0], SEEK_SET)= 0
*read(4*, \0\0\'\n\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\vS|...,
4096) = 4096
*read(4,*
\...@by\233\300\226bg@\342\376\n\341\34...@\266mo\267\241..., 49152) =
49152
*read(4,* ]\3...@\34\21\376\224 +...@g\245\\8\247\3356@6\277\213.\f...,
4096) = 4096
--- SIGSEGV (Segmentation fault) @ 0 (0) ---
+++ killed by SIGSEGV +++
Process 20970 detached
I'm not sure if it faults because it can't read from file handler 4
(Judete.shp), or if it's something else... I also tried by not
specifying Judete in the layer list, but it still dies (while reading
Judete.shp!).
I will look further into problems with AGG (because it renders fine with
PNG)
Thank you for your help - now I know where to look.
Regards,
Adrian
P.S. Ideas are welcome!
Andreas Albarello wrote:
On 16 Jun 2009, at 14:25, Adrian Popa wrote:
Hello David,
I got the 500 errors from the http error log file. Unfortunately no
additional information is provided.
I added the MS_ERRORFILE and DEBUG statements to my map file and
although the errorfile is created, I don't get any other information
(neither when the map loads perfectly, nor when it dies with
premature end of script headers.
I read RFC 28 (http://mapserver.org/development/rfc/ms-rfc-28.html)
and they specified that the server should be compiled with
-DENABLE_STDERR_DEBUG, so I recompiled mapserver:
CFLAGS=-DENABLE_STDERR_DEBUG ./configure --with-freetype --with-png
--with-agg=../agg-2.5 --with-proj --with-ogr --with-gdal --with-xml2
--with-wfs --with-wcs --with-wmsclient --with-wfsclient
