Figure.solar crashes on Linux when drawing nautical terminator for 2024-04-16T06:00:00

[MySlientWind]

Description of the problem

When I try to draw nautical terminator for year=2024, month=4, day=16, hour=6, minute=0, second=0, I got

malloc(): invalid size (unsorted)
Aborted (core dumped)

There is no problem to draw it for other time, such as year=2024, month=4, day=16, hour=6, minute=1, second=0

Minimal Complete Verifiable Example

import datetime
import pygmt

fig = pygmt.Figure()
fig.coast(region="d", projection="W0/15c", land="darkgreen", water="lightblue")

terminator_datetime = datetime.datetime(year=2024,
                                        month=4,
                                        day=16,
                                        hour=6,
                                        minute=0,
                                        second=0)

fig.solar(
    terminator="nautical",
    terminator_datetime=terminator_datetime,
    fill="blue@90",
    pen="0.5p",
)

fig.savefig("demo.png")

Full error message

malloc(): invalid size (unsorted)
Aborted (core dumped)

System information

PyGMT information:
  version: v0.10.0
System information:
  python: 3.12.1 | packaged by conda-forge | (main, Dec 23 2023, 08:03:24) [GCC 12.3.0]
  executable: /home/{my_user_name}/software/anaconda3/envs/pygmt/bin/python
  machine: Linux-4.18.0-477.13.1.el8_8.x86_64-x86_64-with-glibc2.28
Dependency information:
  numpy: 1.26.3
  pandas: 2.1.4
  xarray: 2023.12.0
  netCDF4: 1.6.5
  packaging: 23.2
  contextily: None
  geopandas: None
  IPython: 8.19.0
  rioxarray: None
  ghostscript: 9.56.1
GMT library information:
  binary version: 6.4.0
  cores: 80
  grid layout: rows
  image layout: 
  library path: /work/{my_user_name}/anaconda3/envs/pygmt/lib/libgmt.so
  padding: 2
  plugin dir: /work/{my_user_name}/anaconda3/envs/pygmt/lib/gmt/plugins
  share dir: /home/{my_user_name}/software/anaconda3/envs/pygmt/share/gmt
  version: 6.4.0

[welcome]

👋 Thanks for opening your first issue here! Please make sure you filled out the template with as much detail as possible. You might also want to take a look at our contributing guidelines and code of conduct.

[seisman]

It works well for me, but I'm using macOS. Could you please add verbose="d" to the Figure.solar method and post the debugging messages?

[yvonnefroehlich]

For me, it fails under Linux as reported, but it works under Windows.

Output of verbose="d":

solar [DEBUG]: Map distance calculation will be using great circle approximation with authalic auxiliary latitudes and authalic (R_2) radius = 6371007.1809 m, in meter.
solar [DEBUG]: Map distance calculation will be using great circle approximation with authalic auxiliary latitudes and authalic (R_2) radius = 6371007.1809 m, in meter.
solar [DEBUG]: Reset MAP_ANNOT_OBLIQUE to anywhere
solar [DEBUG]: Projected values in meters: -1.80199e+07 1.80199e+07 -9.00996e+06 9.00996e+06
solar [DEBUG]: Computed automatic parameters using dimension scaling: 0.908088
solar [INFORMATION]: Map scale is 2402.66 km per cm or 1:2.40266e+08.
solar [DEBUG]: Running in PS mode modern
solar [DEBUG]: Use PS filename /home/yfroe/.gmt/sessions/gmt_session.28939/gmt_4.ps-
solar [DEBUG]: Append to hidden PS file /home/yfroe/.gmt/sessions/gmt_session.28939/gmt_4.ps-
solar [DEBUG]: Got session name as pygmt-session and default graphics formats as pdf
solar [DEBUG]: Basemap order: Frame = above  Grid = above  Tick/Annot = above
solar [DEBUG]: gmt_get_smallcircle: Added extra point at 0/-90.9771
solar [DEBUG]: gmt_get_smallcircle: Added extra point at 0/-89.0292
solar [DEBUG]: gmtlib_determine_pole: N = 363 Multiples of 360: 0  Residual: -4.54747e-13 Polygon contains south (CW) pole.
solar [INFORMATION]: Plot small circle with pole at 89.9773/10.0419 and radius 102 degrees
solar [DEBUG]: Polar cap: 1
solar [DEBUG]: Try to include S pole in polar cap path
solar [DEBUG]: West longitude = -180.  East longitude = 180
solar [DEBUG]: gmtlib_determine_pole: N = 363 Multiples of 360: 0  Residual: -4.54747e-13 Polygon contains south (CW) pole.
solar [DEBUG]: First longitude = 89.9773.  Last longitude = 89.9773
solar [DEBUG]: Crossing at -180,-90
solar [DEBUG]: k at point closest to lon -180 is = 0 [n = 363]
solar [DEBUG]: Created path from 180/-90 to 180/-90 [0 points]
solar [DEBUG]: Add perimeter data from k0->n [0->363], then 0->k0 [0]
solar [DEBUG]: Add path from -180/-90 to -180/-90 [0 points]
solar [DEBUG]: New path has 364 points, we allocated 363 points
malloc(): invalid size (unsorted)


Fatal Python error: Aborted

[yvonnefroehlich]

It looks like the issue is related to the fill parameter. If specifying a fill color via fill the code fails, but it works if not using fill or fill=None [Default]. The "problematic" date part seems to be hour=6, minute=0.

import datetime
import pygmt

term_datetime = datetime.datetime(
    year=2024,
    month=4,
    day=16,
    hour=6,
    minute=0,  # fails with a fill color
    # minute=1,  # works with a fill color
    second=0,
)
print(term_datetime)

fig = pygmt.Figure()
fig.coast(region="d", projection="W10c", shorelines=True, frame=True)

fig.solar(
    terminator="nautical",
    terminator_datetime=term_datetime,
    # fill=None,  # works
    fill="blue",  # fails
    # pen="1p,red,solid",  # works
)

fig.show()

[seisman]

The equivalent GMT CLI is:

gmt solar -Rd -JW10c -Baf -Tn+d2024-04-16T06:00:00 -Gblue -png map

Could you please try if it fails for you on Linux?

[yvonnefroehlich]

The equivalent GMT CLI is:

gmt solar -Rd -JW10c -Baf -Tn+d2024-04-16T06:00:00 -Gblue -png map

Could you please try if it fails for you on Linux?

Yes, it fails with GMT 6.4 and 6.4.0_965606f_2022.03.14 on Linux.

(gmt_env_dev) yfroe@gpiseis16:~> gmt solar -Rd -JW10c -Baf -Tn+d2024-04-16T06:00:00 -Gblue -png map
malloc(): invalid size (unsorted)
Aborted (core dumped)
(gmt_env_dev) yfroe@gpiseis16:~> gmt --version
6.4.0_965606f_2022.03.14

[MySlientWind]

Output of verbose="d":

solar [DEBUG]: Map distance calculation will be using great circle approximation with authalic auxiliary latitudes and authalic (R_2) radius = 6371007.1809 m, in meter.
solar [DEBUG]: Map distance calculation will be using great circle approximation with authalic auxiliary latitudes and authalic (R_2) radius = 6371007.1809 m, in meter.
solar [DEBUG]: Reset MAP_ANNOT_OBLIQUE to anywhere
solar [DEBUG]: Projected values in meters: -1.80199e+07 1.80199e+07 -9.00996e+06 9.00996e+06
solar [DEBUG]: Computed automatic parameters using dimension scaling: 0.908088
solar [INFORMATION]: Map scale is 2402.66 km per cm or 1:2.40266e+08.
solar [DEBUG]: Running in PS mode modern
solar [DEBUG]: Use PS filename /home/{my_user_name}/.gmt/sessions/gmt_session.852013/gmt_1.ps-
solar [DEBUG]: Append to hidden PS file /home/{my_user_name}/.gmt/sessions/gmt_session.852013/gmt_1.ps-
solar [DEBUG]: Got session name as pygmt-session and default graphics formats as pdf
solar [DEBUG]: Basemap order: Frame = above  Grid = above  Tick/Annot = above
solar [DEBUG]: gmt_get_smallcircle: Added extra point at 0/-90.9758
solar [DEBUG]: gmt_get_smallcircle: Added extra point at 0/-89.0332
solar [DEBUG]: gmtlib_determine_pole: N = 363 Multiples of 360: 0  Residual: -1.27898e-13 Polygon contains south (CW) pole.
solar [INFORMATION]: Plot small circle with pole at 89.9329/10.3111 and radius 102 degrees
solar [DEBUG]: Polar cap: 1
solar [DEBUG]: Try to include S pole in polar cap path
solar [DEBUG]: West longitude = -180.  East longitude = 180
solar [DEBUG]: gmtlib_determine_pole: N = 363 Multiples of 360: 0  Residual: -1.27898e-13 Polygon contains south (CW) pole.
solar [DEBUG]: First longitude = 89.9329.  Last longitude = 89.9329
solar [DEBUG]: Crossing at -180,-90
solar [DEBUG]: k at point closest to lon -180 is = 0 [n = 363]
solar [DEBUG]: Created path from 180/-90 to 180/-90 [0 points]
solar [DEBUG]: Add perimeter data from k0->n [0->363], then 0->k0 [0]
solar [DEBUG]: Add path from -180/-90 to -180/-90 [0 points]
solar [DEBUG]: New path has 364 points, we allocated 363 points
malloc(): invalid size (unsorted)
Aborted (core dumped)

Output of

gmt solar -Rd -JW10c -Baf -Tn+d2024-04-16T06:00:00 -Gblue -png map

malloc(): invalid size (unsorted)
Aborted (core dumped)

[seisman]

It's likely an upstream bug, although I can't test if it has been fixed in the upcoming GMT 6.5.0 release.

Ping @PaulWessel.

[PaulWessel]

6.5 master (i.e., release) on macOS arm64:

gmt solar -Rd -JW10c -Baf -Tn+d2024-04-16T06:00:00 -Gblue -png map

[yvonnefroehlich]

Thanks @PaulWessel for immediately looking at this. So far, this issue appears to be OS-dependent: it works on Windows and macOS, but it fails on Linux.

[PaulWessel]

So we need someone with Linux to run that command in ddd for example and learn what the crash is and why.

[seisman]

It still crashes with GMT master branch. Trying to debug it, and this is the callstack and shows where it crashes:

[PaulWessel]

So it SEGVs in line 565? That is a calloc call but the message says malloc? What are the values of Belem and size? The message about size is invalid (unsorted) is strange was size is a scalar, no??

[seisman]

nelem=2048 and size=8.

[seisman]

Does this debug message make sense to you?

solar [DEBUG]: New path has 364 points, we allocated 363 points

As the OP says, using a slightly different time works:

$ gmt solar -Rd -JW10c -Tn+d2024-04-16T06:01:00 -Gblue -png map
solar [DEBUG]: gmt_get_smallcircle: Added extra point at 180/-86.5149
solar [DEBUG]: gmt_get_smallcircle: Added extra point at 0/-89.0502
solar [DEBUG]: gmtlib_determine_pole: N = 363 Multiples of 360: 1  Residual: 2.27374e-13 Polygon contains south (CCW) pole.
solar [INFORMATION]: Plot small circle with pole at 89.6829/10.3113 and radius 102 degrees
solar [DEBUG]: Polar cap: 1
solar [DEBUG]: Try to include S pole in polar cap path
solar [DEBUG]: West longitude = -180.  East longitude = 180
solar [DEBUG]: gmtlib_determine_pole: N = 363 Multiples of 360: 1  Residual: 2.27374e-13 Polygon contains south (CCW) pole.
solar [DEBUG]: Make polygon to clockwise
solar [DEBUG]: First longitude = 89.6829.  Last longitude = 89.6829
solar [DEBUG]: Crossing at -180,-86.5149
solar [DEBUG]: k at point closest to lon -180 is = 358 [n = 363]
solar [DEBUG]: Created path from 180/-90 to 180/-86.5149 [48 points]
solar [DEBUG]: Add perimeter data from k0->n [358->363], then 0->k0 [358]
solar [DEBUG]: Add path from -180/-86.5149 to -180/-90 [48 points]
solar [DEBUG]: New path has 459 points, we allocated 459 points

[PaulWessel]

Could you try on Linux to change line 9707 in gmt_plot.c to be

`n_new = 2 * MIN (1, perim_n) + n;`

since perim_n is 0 this is a bit suspect. Does that help?

[PaulWessel]

Sorry, MAX of course

[seisman]

Yes, n_new = 2 * MAX (1, perim_n) + n; works.

[seisman]

The upstream bug has been fixed in the GMT's master branch and the working version GMT 6.5.0 will be released soon.