Puppy Linux Discussion Forum

rockedge wrote: Sun Mar 19, 2023 6:20 pm

I think we are careful, hundreds of hours have been invested in being so.

I didn't mean to offend you. I appreciate the efforts of puppy developers.

All I wish is to be able to use puppy as I have been using it. I was much confused about the behaviors of connman and LADSPA plugin, but not any longer.

Has anyone tied to compile the Broadcom BCM4360 driver and the Broadcom BCM43228 802.11a/b/g/n driver against the F96-CE_3 kernel sources?

@rockedge - linked here is the source I used for my 6.0.x wl drivers (worked despite some errors):
https://www.mediafire.com/file/3ik8rv87 ... 7.zip/file
I can have a go at it later today.

@rockedge - I made a 6.0.12 FP driver with linked source (compiled without errors), ran depmod, but then modprobe failed to insert - 'invalid argument'.
Then I swapped in my 6.1.14ao (aufs/overlayfs) kernel - it's wl driver loaded fine, so looks like kernel issue.

ozsouth wrote: Tue Mar 21, 2023 3:08 am

@rockedge - linked here is the source I used for my 6.0.x wl drivers (worked despite some errors):
https://www.mediafire.com/file/3ik8rv87 ... 7.zip/file
I can have a go at it later today.

Thank you ozsouth.
As with the debian and ubuntu sources I reported earlier in this thread, this source also compiles fine (actually indeed without any warning).
However, like my previous attempts, and as ozsouth also confirmed above, it fails to load with a symbol missmatch error.
This is also the case with another out-of-tree module I tried. Compiles but does not load.
It would appear that the kernel is not made from the provided kernel source or there is a compiler miss-match between the kernel building and the CE2/3 one.

rockedge you might want to compile the CE3 kernel in CE3 if you have not done so originally.

EDIT: So annoying - only 10 broadcom chipsets need the wl driver - the rest can use b43 with firmware.

The following Broadcom chipsets need the wl driver:
BCM4311 , BCM4312 , BCM4321 , BCM4322 , BCM4331 , BCM4352 , BCM4360 , BCM43142 , BCM43227 , BCM43228

In case someone can pinpoint the problem better, here is the kernel "complains" when trying to load the compiled wl module

[   10.786835] wl: disagrees about version of symbol cfg80211_inform_bss_frame_data
[   10.786837] wl: Unknown symbol cfg80211_inform_bss_frame_data (err -22)
[   10.786852] wl: disagrees about version of symbol skb_put
[   10.786853] wl: Unknown symbol skb_put (err -22)
[   10.786855] wl: disagrees about version of symbol cfg80211_scan_done
[   10.786856] wl: Unknown symbol cfg80211_scan_done (err -22)
[   10.786862] wl: disagrees about version of symbol unregister_netdev
[   10.786863] wl: Unknown symbol unregister_netdev (err -22)
[   10.786879] wl: disagrees about version of symbol cfg80211_disconnected
[   10.786880] wl: Unknown symbol cfg80211_disconnected (err -22)
[   10.786882] wl: disagrees about version of symbol ether_setup
[   10.786883] wl: Unknown symbol ether_setup (err -22)
[   10.786886] wl: disagrees about version of symbol wake_up_process
[   10.786886] wl: Unknown symbol wake_up_process (err -22)
[   10.786888] wl: disagrees about version of symbol dev_driver_string
[   10.786889] wl: Unknown symbol dev_driver_string (err -22)
[   10.786891] wl: disagrees about version of symbol dev_addr_mod
[   10.786892] wl: Unknown symbol dev_addr_mod (err -22)
[   10.786893] wl: disagrees about version of symbol eth_type_trans
[   10.786894] wl: Unknown symbol eth_type_trans (err -22)
[   10.786900] wl: disagrees about version of symbol alloc_netdev_mqs
[   10.786901] wl: Unknown symbol alloc_netdev_mqs (err -22)
[   10.786902] wl: disagrees about version of symbol wiphy_new_nm
[   10.786903] wl: Unknown symbol wiphy_new_nm (err -22)
[   10.786910] wl: disagrees about version of symbol wiphy_register
[   10.786911] wl: Unknown symbol wiphy_register (err -22)
[   10.786912] wl: disagrees about version of symbol cfg80211_put_bss
[   10.786913] wl: Unknown symbol cfg80211_put_bss (err -22)
[   10.786915] wl: disagrees about version of symbol cfg80211_roamed
[   10.786916] wl: Unknown symbol cfg80211_roamed (err -22)
[   10.786920] wl: disagrees about version of symbol cfg80211_gtk_rekey_notify
[   10.786921] wl: Unknown symbol cfg80211_gtk_rekey_notify (err -22)
[   10.786924] wl: disagrees about version of symbol ieee80211_get_channel_khz
[   10.786925] wl: Unknown symbol ieee80211_get_channel_khz (err -22)
[   10.786928] wl: disagrees about version of symbol cfg80211_ibss_joined
[   10.786929] wl: Unknown symbol cfg80211_ibss_joined (err -22)
[   10.786931] wl: disagrees about version of symbol skb_pull
[   10.786931] wl: Unknown symbol skb_pull (err -22)
[   10.786938] wl: disagrees about version of symbol cfg80211_michael_mic_failure
[   10.786939] wl: Unknown symbol cfg80211_michael_mic_failure (err -22)
[   10.786940] wl: disagrees about version of symbol kfree_skb_reason
[   10.786941] wl: Unknown symbol kfree_skb_reason (err -22)
[   10.786943] wl: disagrees about version of symbol skb_push
[   10.786944] wl: Unknown symbol skb_push (err -22)
[   10.786949] wl: disagrees about version of symbol register_netdev
[   10.786949] wl: Unknown symbol register_netdev (err -22)
[   10.786953] wl: disagrees about version of symbol wiphy_unregister
[   10.786954] wl: Unknown symbol wiphy_unregister (err -22)
[   10.786960] wl: disagrees about version of symbol cfg80211_get_bss
[   10.786961] wl: Unknown symbol cfg80211_get_bss (err -22)
[   10.786964] wl: disagrees about version of symbol netif_tx_wake_queue
[   10.786965] wl: Unknown symbol netif_tx_wake_queue (err -22)
[   10.786975] wl: disagrees about version of symbol netif_rx
[   10.786976] wl: Unknown symbol netif_rx (err -22)
[   10.786978] wl: disagrees about version of symbol __netdev_alloc_skb
[   10.786979] wl: Unknown symbol __netdev_alloc_skb (err -22)
[   10.786989] wl: disagrees about version of symbol __dev_kfree_skb_any
[   10.786990] wl: Unknown symbol __dev_kfree_skb_any (err -22)
[   10.787000] wl: disagrees about version of symbol skb_clone
[   10.787001] wl: Unknown symbol skb_clone (err -22)
[   10.787004] wl: disagrees about version of symbol cfg80211_report_wowlan_wakeup
[   10.787005] wl: Unknown symbol cfg80211_report_wowlan_wakeup (err -22)
[   10.787012] wl: disagrees about version of symbol cfg80211_inform_bss_data
[   10.787013] wl: Unknown symbol cfg80211_inform_bss_data (err -22)
[   10.787023] wl: disagrees about version of symbol cfg80211_connect_done
[   10.787024] wl: Unknown symbol cfg80211_connect_done (err -22)
[   10.787026] wl: disagrees about version of symbol wiphy_free
[   10.787026] wl: Unknown symbol wiphy_free (err -22)

The miss-match appears to be with the cfg80211 subsystem, but I can not tell the source of it.

Playing around I put "f96-ce" in a Google search and there is a neat little graphic next to it.

retiredt00 wrote: Tue Mar 21, 2023 5:52 am

you might want to compile the CE3 kernel in CE3 if you have not done so originally.

I had some time and I tried it myself
It did not work. The wl driver still failed to load.
Then I noticed the old config in the kernel source SFS and since older versions of the kernel worked, I build the kernel with these setting.
The same driver (that I did not recompile) loaded and worked!
So the problem is the new kernel configuration.
Unfortunately there are many changes between the 2 configuration files so it may take some more digging to figure out.
In a fast look, it appears like that several network related items have been removed from the official CE3 kernel, while the distributed switch architecture infrastructure has been added.
This might be the issue, but at least for now I can use the network with F64-CE3

retiredt00 wrote: Wed Mar 22, 2023 4:26 pm

Then I noticed the old config in the kernel source SFS and since older versions of the kernel worked, I build the kernel with these setting.
The same driver (that I did not recompile) loaded and worked!
So the problem is the new kernel configuration.

Which version of F96-CE had the kernel with the configuration that worked?

Marv wrote: Wed Mar 22, 2023 5:06 pm

retiredt00 wrote: Wed Mar 22, 2023 4:26 pm

Then I noticed the old config in the kernel source SFS and since older versions of the kernel worked, I build the kernel with these setting.
The same driver (that I did not recompile) loaded and worked!
So the problem is the new kernel configuration.

Which version of F96-CE had the kernel with the configuration that worked?

Sorry I am not sure. I think was CE
However the working configuration is in the current kernel_sources_fossapup64_9.6.sfs as the invisible file ".config".
rockedge should know better in which F64 version this config was used.

F96-CE_3 is nice, like most of pups, puppy package manger only allows for a single download each time. I recall some pups batch downloads (i.e. 2 or 3 apps) is possible...
Can this be changed?

I seem to remember PPM got changed to one program selection and download at a time.

Due to the fact that if dependency stuff is also needed to be downloaded, it will be.

Sometimes a different program, that is also selected, will need the same dependency stuff, but not see that it is already downloaded and installed.

Plus, if you have a small save file.

Multiple selected stuff to download, can quickly be more than the save can handle.

There has never been a good warning to tell you the save is full.
What usually happens, is the download and install just does not work, because no room for it in the save.

Also, if you had multiple selected programs to download and install.

If one of them broke the Puppy install.
How are you going to figure out which one it was.

PPM getting software (from non-Puppy specific repositories), that was not specifically compiled for the specific version of Puppy, is not always a perfect 100% fit.
One of the weaknesses of PPM.
Better to download, install, and check program operation, before getting another program.

retiredt00 wrote: Tue Mar 21, 2023 5:52 am

ozsouth wrote: Tue Mar 21, 2023 3:08 am

@rockedge - linked here is the source I used for my 6.0.x wl drivers (worked despite some errors):
https://www.mediafire.com/file/3ik8rv87 ... 7.zip/file
I can have a go at it later today.

Thank you ozsouth.
As with the debian and ubuntu sources I reported earlier in this thread, this source also compiles fine (actually indeed without any warning).
However, like my previous attempts, and as ozsouth also confirmed above, it fails to load with a symbol missmatch error.
This is also the case with another out-of-tree module I tried. Compiles but does not load.
It would appear that the kernel is not made from the provided kernel source or there is a compiler miss-match between the kernel building and the CE2/3 one.

rockedge you might want to compile the CE3 kernel in CE3 if you have not done so originally.

@ozsouth and @retiredt00,

Thank you for the bug reports pertaining to the failed compile/loading of the Broadcom wl driver in F96-CE_3.

Using F96-CE_3 and Woof-CE, I have now generated a new kernel 6.0.12-FP and devx.

Subsequently, I downloaded ozsouth's broadcom-wl source mentioned at the following link:

viewtopic.php?p=84765#p84765

I compiled the wl driver as follows: make, make install, depmod -A, modprobe wl

There were no errors and lsmod confirms the wl driver is successfully loaded.

I don't know if you have limited internet access, but if not, perhaps you could download and test the F96-CE_3_TESTING iso at the following link. The iso contains the following updated files: vmlinuz, zdrv_fossapup64_9.6.sfs, kernel_sources_fossapup64_9.6.sfs, and devx_fossapup64_9.6.sfs.

https://www.smokey01.com/radky/Testing/ ... ESTING.iso

Thank you.

---

# pwd
/root/Downloads/broadcom-wl-patch-linux4.7
#
# make
KBUILD_NOPEDANTIC=1 make -C /lib/modules/`uname -r`/build M=`pwd`
make[1]: Entering directory '/usr/src/linux'
CFG80211 API is prefered for this kernel version
Using CFG80211 API
  CC [M]  /root/Downloads/broadcom-wl-patch-linux4.7/src/shared/linux_osl.o
  CC [M]  /root/Downloads/broadcom-wl-patch-linux4.7/src/wl/sys/wl_linux.o
  CC [M]  /root/Downloads/broadcom-wl-patch-linux4.7/src/wl/sys/wl_iw.o
  CC [M]  /root/Downloads/broadcom-wl-patch-linux4.7/src/wl/sys/wl_cfg80211_hybrid.o
  LD [M]  /root/Downloads/broadcom-wl-patch-linux4.7/wl.o
CFG80211 API is prefered for this kernel version
Using CFG80211 API
  MODPOST /root/Downloads/broadcom-wl-patch-linux4.7/Module.symvers
  CC [M]  /root/Downloads/broadcom-wl-patch-linux4.7/wl.mod.o
  LD [M]  /root/Downloads/broadcom-wl-patch-linux4.7/wl.ko
make[1]: Leaving directory '/usr/src/linux'
# 
# make install
install -D -m 755 wl.ko /lib/modules/`uname -r`/kernel/drivers/net/wireless
# 
# depmod -A
# 
# modprobe wl
#
# lsmod
Module                  Size  Used by
wl                   6381568  0

radky wrote: Thu Mar 23, 2023 1:30 pm

Using F96-CE_3 and Woof-CE, I have now generated a new kernel 6.0.12-FP and devx.

Subsequently, I downloaded ozsouth's broadcom-wl source mentioned at the following link:

viewtopic.php?p=84765#p84765

I compiled the wl driver as follows: make, make install, depmod -A, modprobe wl

There were no errors and lsmod confirms the wl driver is successfully loaded.

I don't know if you have limited internet access, but if not, perhaps you could download and test the F96-CE_3_TESTING iso at the following link. The iso contains the following updated files: vmlinuz, zdrv_fossapup64_9.6.sfs, kernel_sources_fossapup64_9.6.sfs, and devx_fossapup64_9.6.sfs.

https://www.smokey01.com/radky/Testing/ ... ESTING.iso

Thank you.

Other than the wl driver compile related changes, are there any other aspects of the rebuilt F96-CE_3_Testing kernel that should be looked at. I'm running it in both the 'Testing' F96-CE_3 and in upup-CE_D1 (paths adjusted, overlayfs) at the moment.

Marv wrote: Thu Mar 23, 2023 7:29 pm

Other than the wl driver compile related changes, are there any other aspects of the rebuilt F96-CE_3_Testing kernel that should be looked at. I'm running it in both the 'Testing' F96-CE_3 and in upup-CE_D1 (paths adjusted, overlayfs) at the moment.

@Marv

In F96-CE_2 the 6.0.12 kernel was adjusted for improved support of IPv6, and in F96-CE_3_TESTING there are additional adjustments to preclude symbol mismatch errors while compiling.

Hopefully, F96-CE_3_TESTING will address recent bug reports by retired00 and ozsouth when compiling the Broadcom wl driver, but we wait for their reports.

Otherwise, at the moment I'm not aware of other functional issues with the 6.0.12-FP kernel. Of course, the current F96 zdrv is of modest size with relatively broad hardware support, but it is not a mega zdrv with universal built-in support for every possible hardware iteration.

Note: F96-CE_4 will include the kernel adjustments, plus updates for the main Puppy SFS.

radky wrote: Thu Mar 23, 2023 1:30 pm

There were no errors and lsmod confirms the wl driver is successfully loaded.

I don't know if you have limited internet access, but if not, perhaps you could download and test the F96-CE_3_TESTING iso at the following link. The iso contains the following updated files: vmlinuz, zdrv_fossapup64_9.6.sfs, kernel_sources_fossapup64_9.6.sfs, and devx_fossapup64_9.6.sfs.

https://www.smokey01.com/radky/Testing/ ... ESTING.iso

Thank you.

Dear rockedge
In a very fast test (replacing vmlinuz and zdrv from the test ISO to the old installation) the wl driver failed to load and had the the same unknown symbol errors as with the kernels for CE2 and CE3..
This was the driver that is working OK with the CE1 kernel configuration.
As far as I can see the configuration changes are minimal, so this should not be surprising.
BTW you should notice that the driver fails when is trying to cooporate with the cfg80211 subsystem (if is not used, does not show any problem).
So I think that there is something in the cfg80211 subsystem configuration and/or modules that is off or missing that triggers the breakage
Of cource I should compile the drivers with the new devx and source to validate or refute the above fast test.
Hopefully I'll manage at some point sooner rather than later

retiredt00 wrote: Thu Mar 23, 2023 9:43 pm

Dear rockedge
In a very fast test (replacing vmlinuz and zdrv from the test ISO to the old installation) the wl driver failed to load and had the the same unknown symbol errors as with the kernels for CE2 and CE3..
This was the driver that is working OK with the CE1 kernel configuration.
As far as I can see the configuration changes are minimal, so this should not be surprising.
BTW you should notice that the driver fails when is trying to cooporate with the cfg80211 subsystem (if is not used, does not show any problem).
So I think that there is something in the cfg80211 subsystem configuration and/or modules that is off or missing that triggers the breakage
Of course I should compile the drivers with the new devx and source to validate or refute the above fast test.
Hopefully I'll manage at some point sooner rather than later

retiredt00

F96-CE_3_TESTING contains 4 updated files:

1- vmlinuz
2-zdrv_fossapup64_9.6.sfs,
3-kernel_sources_fossapup64_9.6.sfs
4-devx_fossapup64_9.6.sfs.

As time permits, it would be best to test compiling in the F96-CE_3_TESTING iso, rather than compiling with a hybrid mix of files from F96-CE_3 and F96-CE_3_TESTING.

Thanks

@radky @retiredt00 - I confirm that under F96CE-3T, wl compiles without error, loads, & is present via lsmod. Will now fire up old clunker with bcm4360 chip & do actual internet test of wl driver & report back.
LATER: wl driver works with bcm4360. Posting from it now. b43, bcma, brcmsmac & ssb needed to be blacklisted.

ozsouth wrote: Thu Mar 23, 2023 11:13 pm

@radky @retiredt00 - I confirm that under F96CE-3T, wl compiles without error, loads, & is present via lsmod. Will now fire up old clunker with bcm4360 chip & do actual internet test of wl driver & report back.

LATER: wl driver works with bcm4360. Posting from it now. b43, bcma, brcmsmac & ssb needed to be blacklisted.

@ozsouth

Thanks for evaluating the F96-CE_3_TESTING iso and confirming successful compiling of the Broadcom wl driver.

@radky Booted F96-CE_3-Test and loaded the Devx, kernel sources.

Downloaded VirtualBox v7.0.6 as a .deb package and installed it. Then ran /sbin/vboxconfig which completed successfully and the VIrtualBox service daemon was restarted. The download was from the VirtualBox official website.

So the VIrtualBox virtual disk driver was successfully compiled.

DELL VOSTRO 15 laptop as host machine. Outfitted with a single core Intel celeron CPU and 4 gb of RAM.

retiredt00 wrote: Thu Mar 23, 2023 9:43 pm

Of cource I should compile the drivers with the new devx and source to validate or refute the above fast test.

I did a fresh install from the test ISO and compiled the broadcom-sta module from the latest debian testing branch and well as an out of tree broadcom webcam module from a github source.
The wl driver compiled with some warnings about depreciated paths and options while the webcam driver complained about missing system map during installation.
However, the both loaded and worked as expected.
So I think this kernel/devx combination is good.
Thank you.

Started F96-CE_3-Test on a DELL PowerEdge server, loaded the Devx and kernel sources using SFS-Load.

Performed the same test of downloading a fresh VirtualBox 7.0.6 as a .deb for Ubuntu Focal. Installed with left-click in a Rox-filer window.

Vbox started smoothly and set up of a KLV VM went like a textbook exercise.

Then with /sbin/vboxconfig the new VBox disk driver was successfully built AFTER creating the new user:group vboxusers

Thanks to all who evaluated the transitional F96-CE_3_TESTING iso.

F96-CE_3_TESTING.iso is now removed

MochiMoppel wrote:

...F96-CE-3 wouldn't detect my USB scanner.

Unfortunately Xsane appears to be not installed at all, though the xsaneshell script exists. Will it be included in the ISO some day?

@MochiMoppel

For those with scanners, the information below may be helpful.

F96 has the following (ubuntu-focal) sane/xsane packages installed:

1 - libsane_1.0.29 : [ubuntu-focal-main] (ALREADY INSTALLED) API library for scanners
2 - libsane-common_1.0.29 : [ubuntu-focal-main] (ALREADY INSTALLED) API library for scanners -- documentation and support files
3 - sane-utils_1.0.29 : [ubuntu-focal-main] (ALREADY INSTALLED) API library for scanners -- utilities
4 - xsane_0.999 : [ubuntu-focal-universe] (ALREADY INSTALLED) featureful graphical frontend for SANE Scanner Access Now Easy
5 - xsane-common_0.999 : [ubuntu-focal-universe] (ALREADY INSTALLED) xsane architecture independent files

---

XSANE is the frontend for SANE scanner drivers.

Each new release of SANE provides driver support for additional scanners.

For sane 1.0.29 in F96, the following information may be helpful.

Scanners supported by SANE-1.0.29:
http://www.sane-project.org/sane-backends-1.0.29.html

SANE Project:
http://www.sane-project.org/

@MochiMoppel

I have a very old USB connected HP All-in-1 printer/scanner. XSane works perfectly in F96-3 and detects my scanner after loading hplip_3.20.3+dfsg0-2 from the package manager.

radky wrote: Fri Mar 24, 2023 8:52 pm

MochiMoppel wrote:

...F96-CE-3 wouldn't detect my USB scanner.
Unfortunately Xsane appears to be not installed at all, though the xsaneshell script exists. Will it be included in the ISO some day?

@radky Thank you for looking into my scanner issue.
Unfortunately the quote is out of context and misleading. Here is what I wrote in the Vanilla Dpup thread. It should be clear that "not installed" refers to Vanilla Dpup and not F96-CE.

Scanners supported by SANE-1.0.29:
http://www.sane-project.org/sane-backends-1.0.29.html

My scanner is a Microtek SlimScan C6 and according to the list it is supported:

Starting Xsane from the F96-CE menu finds no scanner.
However running sane-find-scanner seems to detect the scanner (see uncommented line). The USB ID matches the one in the SANE compatibility list

# /usr/bin/sane-find-scanner

  # sane-find-scanner will now attempt to detect your scanner. If the
  # result is different from what you expected, first make sure your
  # scanner is powered up and properly connected to your computer.

  # No SCSI scanners found. If you expected something different, make sure that
  # you have loaded a kernel SCSI driver for your SCSI adapter.
  # Also you need support for SCSI Generic (sg) in your operating system.
  # If using Linux, try "modprobe sg".

found USB scanner (vendor=0x05da, product=0x009a) at libusb:004:002
  # Your USB scanner was (probably) detected. It may or may not be supported by
  # SANE. Try scanimage -L and read the backend's manpage

The scanner works fine in Slacko 5.6. The difference: Slacko uses xsane 0.997 while F96 uses xsane 0.999
Any ideas how to fix it?

RE: Microtek SlimScan C6 USB scanner

@MochiMoppel

For Slacko-5.6 and F96 I see the following specs for xsane/sane:

Slacko-5.6:
xsane 0.997
sane 1.0.21 (libs in /usr/lib/sane)

F96:
xsane 0.999
sane 1.0.29 (libs in /usr/lib/x86_64-linux-gnu/sane)

---

Maybe an experienced Microtek user will provide a proper fix for you, but until then you could try the following:

In the working Slacko-5.6 and non-working F96, you could run scanimage -L (with a debug flag) to see if there are any useful output differences or error messages when searching, opening or initializing the microtek backend:

SANE_DEBUG_DLL=3 scanimage -L

RE: Microtek SlimScan C6 USB scanner

@MochiMoppel

There are specific udev rules that may be helpful for some USB devices.

For example, you could create the following file (udev rule) for libsane:

/etc/udev/rules.d/40-libsane.rules

then add the following content:

# Microtek SlimScan C6
ATTRS{idVendor}=="05da", ATTRS{idProduct}=="009a", ENV{libsane_matched}="yes"

then reboot

Hello everybody:

We (@nilsonmorales and me) are working with F96-CE_3 (as the most updated puppy) using xfce and fusilli to create a more visually puppy.

Nilson had made a really good work, but now we found a difficulty that is possible you can help us to solve.

The third beta, works really well, but we have a problem when saving.

I run this modded distro from a USB and I have the savefile.4fs at this USB. When I save using the option flash2save (I have an icon on my desktop), there is NO PROBLEM, so for me is fantastic, because I usually don't shut down the PC. But when I restart or shutdown, an inconvenient appears.

Puppy gives us the option to save or not, but we can't choose one or another option, because the keyboard doesn't allow us to move the pointer and choose one or another option.

My personal option, and the easier, I think, could be eliminate this option to save, or at least wait only 5 seconds instead of the 60 by default.

What file must we edit -- and how--- to do this?

n.b.0:- If I put the save file in a partition, instead of the USB, there is no problem, it automatically saves the sessions without asking you about it, and flash2save doesn't apperar on screen, but as I told u, the problem appears when the savefile on USB.

n.b.- I tried to do this with graffic options (as you can see in the image, unmarking the option SAVE WHEN SHUTDOWN) that different programs give us, but they didn't work at all

THANKS IN ADVANCE for your time, support and interest.- THANKS

@rockedge