XPost: linux.debian.bugs.dist 
 From: zeha@debian.org 
  
 Hi, 
  
 On Sun, Sep 07, 2025 at 12:19:58AM +0200, Philip Hands wrote: 
 >  writes: 
 > AFAICS $TMPFILE (configured to be tmpkbd.*) is being used in the (outer) 
 > setupcon script in order to record the commands used in the (initrd) 
 > setupcon that is being written to "$setupdir"/bin/setupcon 
  
 Yes. 
  
 > $TMPFILE is then used to store a list of things that will need to be 
 > copied to the initrd, and that list is then used to create 
 > "$setupdir"/morefiles, but $TMPFILE looks like it ought to be in either 
 > /run or /tmp, and those should be removed by the trap, so if it's ending 
 > up under $setupdir, what is copying it there? 
  
 From my analysis, setupcon has a code path, which: 
  
 - only runs if no /etc/console-setup/cached* exist 
 - and KMAP is not set in config 
  
 Then it proceeds to use a tempfile (created in /tmp) and call 
 loadkeys with this. 
 Example commands it wants to use for me: 
    kbd_mode -a 
    loadkeys /tmp/tmpkbd.KLgbbu > /dev/null 
  
 setupcon with --setup-dir at the very end then analyzed all commands 
 it wants to use, and tries to *copy* all mentioned filenames into 
 the initramfs. 
 This happens around line 1371-1376 and in the fileargs function. 
  
 > Your patch seems to assume that it is being copied in fileargs(), which 
 > seems to imply that the tmpkbd.* filename is being passed as an argument 
 > to one of the run() calls (if I'm reading the script correctly), but I 
 > don't see where that's happening, and I don't yet understand why it's 
 > happening. 
  
 Indeed this is what happens. I think it only happens as I described 
 above. It's this if-branch: 
 https://salsa.debian.org/installer-team/console-setup/-/blob/a7b 
 a22f15d7927921beef4858e95b4f7f885307/setupcon#L1261 
  
 > If that _is_ what's happening, and if there's a good reason for it, then 
 > I would think that the right place to fix the nondeterminism would be in 
 > the run() call that's causing the copy to be made, rather than by 
 > modifying the filename in fileargs(), [..] 
  
 Agreed. 
  
 > > Now, this script ('/bin/setupcon') is used in the initrd by the 'keymap' 
 initrd script [2]. 
 > > In fact the 'keymap' hook script [3] is what is calling the setupcon 
 script with the --setup-dir option. 
 > > 
 > > In my use case, this issue surfaced due to the use of the cr 
 ptsetup-initramfs package which triggers the run of the 'keymap' hook script 
 by setting 'KEYMAP=y' here[4]. 
 > > 
 > > There *maybe*(speculating, not sure) more packages which triggers this 
 'keymap' hook script. 
 > > 
 > > Apologies for this long answer. I hope this gives some context. 
 > 
 > No, that's great, thanks -- anyway, my reply is longer :-/ 
 > 
 > The question that's still bothering me is: 
 > 
 >   Does anything within the resulting initrd actually make use of the 
 >   'keymap.*' file, and if so, where? 
  
 I'm not sure which file you mean now with 'keymap.*'. But all 
 commands that setupcon would run are also copied into the 
 built-setupcon (including some patching of referenced filenames). 
  
 Chris 
  
 --- SoupGate-Win32 v1.05 
  * Origin: you cannot sedate... all the things you hate (1:229/2)