Opened 9 years ago
Closed 6 years ago
#83 closed defect (wontfix)
ash: crash when using BEGINLIBPATH
Reported by: | dmik | Owned by: | |
---|---|---|---|
Priority: | major | Milestone: | |
Component: | ash | Version: | |
Severity: | medium | Keywords: | |
Cc: |
Description
Given the following script t.sh
:
BEGINLIBPATH='foo' t2.sh
and the empty script t2.sh
an attempt to run
sh t.sh
will result into the following crash:
LIBC PANIC!! _um_free_maybe_lock: Tried to free block twice - block=00032a54 lock=0x1 pid=0x1ef1 ppid=0x1ef0 tid=0x0001 slot=0x00da pri=0x0200 mc=0x0000 ps=0x0010 C:\USR\BIN\SH.EXE Process dumping was disabled, use DUMPPROC / PROCDUMP to enable it. [1] Killed BEGINLIBPATH="fo...
It in particular hurts the install target of python when the plat-os2knix
directory is missing and it tries to create it out of the plat-generic
template.
Change History (4)
comment:1 by , 9 years ago
Component: | → ash |
---|
comment:2 by , 9 years ago
comment:3 by , 9 years ago
Another problem with BEGINLIBPATH is that if it starts with ./
or ../
it simply does not work (DLLs are not found; this affects the icu
package). Perhaps, both problems have the common root. A workaround is to make all the path entries in BEGINLIBPATH absolute but that's not always possible so we should fix this ASAP (or fix the remaining dash bugs and switch to it).
comment:4 by , 6 years ago
Resolution: | → wontfix |
---|---|
Status: | new → closed |
Given that we added BEGINLIBPATH support to dash
within #161, this ticket is no more relevant. The lack of BEGINLIBPATH and friends support in dash was the last missing thing there. Now we don't need ash anymore and won't fix it.
Note that the problem from comment:3 is also solved in dash
: such cases work perfectly there now.
Note that
dash
does not crash in such a case.