Web lists-archives.com

Re: Presence of cygwin64 elements in an existing Cygwin 32 install after updating

On 29/08/2017 23:03, Eliot Moss wrote:
On 8/29/2017 4:57 PM, Keith Christian wrote:
More program failures.  Have rebooted and allowed system to preform rebase:

$ man httping
       0 [main] man 7688 child_info_fork::abort:
C:\cygwin\bin\cygz.dll: Loaded to different address: parent(0x360000)
!= child(0x3E0000)
man: fork failed: Resource temporarily unavailable

Why are all of these programs failing in this way?

Sometimes I have needed to rebase all programs.  To do this,
do the command: rebase-trigger full
If you're not able to run bash to run this, you can run dash
from Windows CMD and give an explicit path, e.g.,

/cygdrive/c/cygwin/bin/rebase-trigger full

would do the right thin on my installation.  But even that won't
help if you're running 32-bit and have installed too many packages.
There is only so much room in the address space for the various
possible DLLs that might be loaded.

in addition to Elliot's comment, you can use

  rebase -si

output to  provide an overview of how many dll's are present
and in which address they are mapped.

  $ rebase -si |sort -r --key=5 | head

will provide the list of the one absorbing more address space.
Except the octave-tisean patological case, that I recommend to not install if you don't really need it, usually LLVM and ICU
several DLL's are the largest:

 $ rebase -si |sort -r --key=5 | head |awk '{print $5,$1}'
0x020b3000 /usr/bin/cygLLVM-4.0.dll
0x01f75000 /usr/bin/cygQt5WebKit-5.dll
0x01def000 /usr/bin/cygLLVM-3.9.dll
0x01af3000 /usr/bin/cygLLVM-3.8.dll
0x0191d000 /usr/bin/cygicudata59.dll
0x0190a000 /usr/bin/cygicudata58.dll
0x01885000 /usr/bin/cygicudata57.dll
0x017ed000 /usr/bin/cygicudata56.dll
0x01239000 /usr/bin/cyggcj-16.dll
0x00f4f000 /usr/bin/cygoctave-4.dll


Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple