LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
syphyr has quit [Quit: Leaving]
tsal has quit [Ping timeout: 268 seconds]
<LEBot>
[slack] <chewitt> depends on what you regard as kinks .. for my use-cases it’s been great for more than a year
tsal has joined #libreelec
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
ngc0202 has quit []
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
buzzmarshall has quit [Quit: Konversation terminated!]
<narf>
10.0.1 fixed a bunch of issues for me
<narf>
on my pi4
jernej has joined #libreelec
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
gouchi has joined #libreelec
<vudiq>
guys, rudi - the update of mtools to 4.0.29 breaks the build when SYSTEM_SIZE>=1024 (i tested with 1024, 2048 - mcopy/mmd fail to copy files/create folders to/in image, 768 works)
<vudiq>
I guess the failure can be avoided by setting env MTOOLS_SKIP_CHECK to 1, the error refers to missing information about number of tracks/heads / being zero
<vudiq>
The devil is in the details: zero number of heads or sectors
<vudiq>
the build finishes, but the files/folders are not created in the image
narf has quit [Read error: Connection reset by peer]
<vpeter>
I did quick look few weeks back but then forgot on it.
<vpeter>
Also alias commands could be different too.
<vpeter>
But even with MTOOLS_SKIP_CHECK it should make valid image. Doesn't it?
chewitt has joined #libreelec
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
<vudiq>
vpeter: I will try with that env and let you know, just setting it up. anyway - what would you suggest to change, because the offset is calculated right i guess. increase number of tracks (cylinders)?
<vpeter>
I think those numbers should be increased. But not blindly: at the end they have to match the size.
narf4 is now known as narf
<vudiq>
vpeter: so e.g. check if the SYSTEM_SIZE is multiple of 128 `[ ${SYSTEM_SIZE} -eq $(( $(( ${SYSTEM_SIZE} / 128 )) * 128 )) ]` and then set number of TRACKS as `$(( ${SYSTEM_SIZE} / 128 ))` - does that look sane to you?
<vudiq>
sorry, not TRACKS, but HEADS
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
<Bardon>
Hello, sorry for being slightly off-topic, I don't know where to ask. I have LE on an RPi. My TV's remote control has "cable" "dvd" "stb" and "vcr" buttons so I'm wondering if I can control LE from my TV's remote
<Bardon>
I'm not sure I can make it work since I need to type a brand code, which probably doesn't apply here
<LEBot>
[slack] <klojum> Are you using the RPi's CEC ? Perhaps have a look at the Keymap Editor addon. Also with enabling debugging the 'alternative' remote keys might show up with their keycodes.
svetlemodry has quit [Quit: Connection reset by beer]
svetlemodry has joined #libreelec
svetlemodry has joined #libreelec
svetlemodry has quit [Changing host]
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
<vudiq>
is there a guideline that is not allowed to have files in a package that contain a space in their filename? the sha256sum / stamp has unwanted stderr output. adding quotes solves it - `data="$(find -L ${stamp} -type f -not -name '.*' 2>/dev/null | sed "s|^${ROOT}/|\"|" | sed "s|$|\"|" | LC_ALL=C sort -u | xargs sha256sum)"
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
jernej has quit [Ping timeout: 268 seconds]
<vudiq>
vpeter: MTOOLS_SKIP_CHECK produces readable image. and I think it is sufficient just to increase number of tracks (make them "denser").
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
keypusher has quit [Ping timeout: 250 seconds]
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
<vpeter>
vudiq: Seems this is preferred solution all over Internet.
<vpeter>
But would be good to understand the issue better.
<vudiq>
vpeter: I don't get it either - the total number of sectors does not change, with TRACKS=32 and SYSTEM_SIZE=2048 the SECTORS (per track) increases from 8192 to 32768, which is still a 2 byte value, but... since the highest bit is set, it is considered a negative value? I did not check the mtools code very deeply
<vudiq>
with TRACKS=SYSTEM_SIZE/16 SECTORS (per track) remains 8192
<vudiq>
but what i can recall is, that it checks for value > 0
falk689 has quit [Remote host closed the connection]
falk689 has joined #libreelec
<vudiq>
vpeter: btw, according to NEWS file, since 4.0.19 mformat should determine the geometry by itself if not supplied / cannot be queried from device
keypusher has joined #libreelec
<vudiq>
also strange: the check should only be performed when media == 0xf0 || media >= 0x100, for hdd the media type is 0xf8 (but I might be mixing things up here)
<vpeter>
Let me try few things.
arch3r has quit [Quit: bye]
arch3r has joined #libreelec
<vudiq>
vpeter: i changed the alias for mformat (removet tracks heads sectors) and it looks like it works
<vudiq>
s/removet/removed
<vudiq>
actually not, fsck fails
<vudiq>
Failed to read sector 4265855.
<vpeter>
My small script seems to be working.
<vpeter>
It works with fsck.fat 4.1 but not with 4.2.
falk689 has quit [Remote host closed the connection]
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
falk689 has joined #libreelec
<vpeter>
Probably mformat uses whole disk image - doesn't know it is second partition too.
<vpeter>
Maybe partition should be formatted as separate partition image.
<vudiq>
that is also an option... I tested my changes (pr #5897) with 512, 1024, 1536, 2048, 2560, 3072 and all works
<vpeter>
And what exactly this new TRACKS calculation do?
narf has quit [Quit: bye]
<vudiq>
vpeter: calculates more accurate number of tracks per one side of the platter based on the system size
<vudiq>
intead of having fixed number of tracks (32) it increases/decreases the number of tracks
<vpeter>
Which mtools version you are using?
<vudiq>
vpeter: the one in the tree
<vudiq>
in master
<vpeter>
Try with 4.0.35.
<vudiq>
and no other changes to TRACKS?
<vpeter>
And with your PR.
narf has joined #libreelec
<vpeter>
Or even without you PR with default 512MB system size.
<vudiq>
ok, I'll test all cases
<vpeter>
Maybe I messed up something here and would like to be sure :)
<vudiq>
bump mtools to .35 + system size 2048 + "tracks fix" -> fsck fail
<vpeter>
So it is not just me :(
<vpeter>
Try with 512.
<vudiq>
bump mtools to .35 + system size 2048 + no "tracks fix" -> devil notice/empty image
<vudiq>
now 512 - does not matter if with/without tracks fix - is the same number of tracks
<vudiq>
btw - bump mtools to .35 + system size 2048 + no "tracks fix" -> devil notice/empty image + fsck fail (!)
<vpeter>
Yes, seems something is still wrong.
<vudiq>
and 512 fails with fsck
<vpeter>
I think part1 should be created, extracted, formatted, copy files on it, checked, merged back.
<vpeter>
And using mformat without heads, cylinders or whatever :)
<vudiq>
should I try here?
<vpeter>
If you want.
<vpeter>
Everything done on part1 should be done on part1.fat. Then mtoools doesn't need offset either.
<vpeter>
But seems it doesn't work this way either.
<vpeter>
Looks like mtools 4.0.35 and fsck.fat 4.2 are not friends!
<vudiq>
so... MTOOLS_SKIP_CHECK :) or how the env is called
<vudiq>
i'll try, but probably fsck will fail
<vudiq>
yup, it does fail on fsck
<vudiq>
Failed to read sector 1116159. (SYSTEM_SIZE=512)
<vpeter>
Will look more tomorrow.
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
<vpeter>
No, all good if using extracted partition. Extract part1, use mformat, copy files, check. All good.
<vudiq>
great
<vpeter>
I think this is safest way. We already extract part1 for fsck.
<vpeter>
So at the end only some small changes are needed (moving code).
<vudiq>
that should be kept if the part1 did not get damaged while inserting part2 back
<vpeter>
I think both partitions are already handled correctly for extract/merge.
<vudiq>
I was also thinking about using `mount -o offset=$(( ${SYSTEM_PART_START} * 512 )) some/temp/folder` and then just copy files unmount
<vpeter>
mount needs root - that's the whole point of using mtools.
<vudiq>
ah, right
<vudiq>
and mformat was with geometry switches or without them?
<vudiq>
when you had "all good"
<vpeter>
Without geometry. But using extracted partition not disk image with offset.
<vpeter>
Maybe without mformat mkfs.vfat could be used.
narmstrong has quit [Remote host closed the connection]
<vudiq>
testing locally
<vudiq>
I get `syslinux.mtools: invalid media signature (not an FAT/NTFS volume?)`
TomTom has quit [Remote host closed the connection]
tortoise has quit [Remote host closed the connection]
<vudiq>
I forgot to swap DISK with LE_TMP/part1.fat32 :D and remove --offset
sopparus_ has quit [Ping timeout: 246 seconds]
falk689 has quit [Quit: No Ping reply in 180 seconds.]
<vpeter>
Seems so. I made 6GB partition and I could mount it.
<vudiq>
I still get 'Disk full' warning when mcopy system, maybe some error in the math
<vudiq>
okay I see... I used OFFSET instead of SYSTEM_PART_START
<vpeter>
You don't need new math. Extracting part1 is already done.
<vudiq>
/o\
<vudiq>
double face palm
<vudiq>
I forgot that bs=512 will be used to multiply SYSTEM_PART_START ...
falk689 has quit [Remote host closed the connection]
<vpeter>
Yes, need some extra space in head for such things.
falk689 has joined #libreelec
<vudiq>
and when kids are running around :D
<vudiq>
okay, got it working
<vudiq>
but I should probably test it with u-boot and bcm2835 as well
<vudiq>
mounted via -o offset, checked md5 for KERNEL and SYSTEM - seems to work
<vudiq>
so I will close my pr...
<vpeter>
No.
<vpeter>
Just change it with this changes.
<vudiq>
I will open new one as the concept is completely different
<vudiq>
vpeter: what is your github ee.. name? so I can put you into authors
<vpeter>
no need
<vpeter>
take the glory :)
<vudiq>
haha :D allright :D
<vudiq>
and you mean the change with extract part1.fat, format, copy, fsck, merge, right?
<vpeter>
Whatever you choose. Simplest way is my small change. But we already do partition extracting so everything could be done on partition file itself.
syphyr has joined #libreelec
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
<vudiq>
vpeter: okay, I added also these error redirects, as otherwise they are just ignored - ` >"${SAVE_ERROR}" 2>&1 || show_error` - e.g. when "Disk full" or other error
tortoise has joined #libreelec
TomTom has joined #libreelec
narmstrong has joined #libreelec
nezach has joined #libreelec
sopparus_ has joined #libreelec
buzzmarshall has joined #libreelec
LEBot has quit [Remote host closed the connection]
<vudiq>
hmm... the OVA - yes - now it is not needed to use the full path as the alias is for the part1.fat file
<vudiq>
I will add that - thanks
<vudiq>
I do not see the benefit of the alias for syslinux.mtools, as it is called only once
<vudiq>
and I updated mtools to latest (.36, tested)
<vpeter>
Seems same.
<vpeter>
But now that you changed every mcopy line you could remove aliases above and use -i directly.
<vpeter>
Also for mformat and mmd.
<vudiq>
and in case of OVA -o switch is needed - now it hangs - wants to confirm overwrite, stdout is redirected, ctrl-c does nothing, kill -SIGTERM also :D
<vudiq>
I would keep the aliases for mcopy, mmd - they are used quite often, in case of a change you update only the alias
<vudiq>
mformat is used only once, so here it makes sense to remove the alias
<vudiq>
had to take out the big guns (-9)
<vpeter>
For -o put it also in alias? It can't hurt.
syphyr has quit [Quit: Leaving]
<vudiq>
I probably have to use single quotes - alias='mcopy -i "${LE_TMP}/part1.fat" -o' - or we can be sure that mktemp does not return path with spaces?
syphyr has joined #libreelec
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
<vudiq>
vpeter: thank you for your help, have a great evening/rest of your day :)
<vudiq>
vpeter: do you know if there is some kind of a guideline, that filenames with spaces are not allowed in packages? I know that path to the ROOT must not include spaces. in one custom package I have files that have spaces, so sha256sum prints some garbage during calculation of the stamp. I can fix that (adding quotes with the sed), but if there is a guideline not to use spaces, I can just rename the files
<vudiq>
:)
nezach has quit [Ping timeout: 268 seconds]
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
gouchi has quit [Remote host closed the connection]
LEBot has quit [Remote host closed the connection]
LEBot has joined #libreelec
LEBot has quit [Remote host closed the connection]