SliTaz Forum
You are not logged in.
Salve, facendo delle prove su di un tablet con processore intel baytrail, mi sono accorto che la versione a 64bit non rileva la memoria (mmc).
Mi piacerebbe avere entrambi funzionanti se fosse possibile.
Paragonando i moduli built-in dei due kernel mi appare questo
tux@tux-baytrail:~/Documenti$ diff -u modules.builtin32 modules.builtin64
--- modules.builtin32 2023-12-01 20:59:04.000000000 +0100
+++ modules.builtin64 2023-12-16 15:43:25.598722000 +0100
@@ -1,4 +1,3 @@
-kernel/arch/x86/kernel/apm.ko
kernel/arch/x86/kernel/iosf_mbi.ko
kernel/arch/x86/kernel/cpu/microcode/microcode.ko
kernel/kernel/configs.ko
@@ -99,7 +98,6 @@
kernel/drivers/ata/pata_piccolo.ko
kernel/drivers/ata/pata_via.ko
kernel/drivers/ata/pata_cmd640.ko
-kernel/drivers/ata/pata_isapnp.ko
kernel/drivers/ata/pata_mpiix.ko
kernel/drivers/ata/pata_acpi.ko
kernel/drivers/ata/ata_generic.ko
@@ -111,12 +109,10 @@
kernel/drivers/block/nvme.ko
kernel/drivers/block/mtip32xx/mtip32xx.ko
kernel/drivers/cdrom/cdrom.ko
-kernel/drivers/char/virtio_console.ko
kernel/drivers/char/nvram.ko
kernel/drivers/char/hw_random/rng-core.ko
kernel/drivers/char/hw_random/intel-rng.ko
kernel/drivers/char/hw_random/amd-rng.ko
-kernel/drivers/char/hw_random/geode-rng.ko
kernel/drivers/char/hw_random/via-rng.ko
kernel/drivers/cpufreq/cpufreq_stats.ko
kernel/drivers/cpufreq/cpufreq_performance.ko
@@ -157,6 +153,7 @@
kernel/drivers/hid/hid-zpff.ko
kernel/drivers/hid/hid-wiimote.ko
kernel/drivers/hid/usbhid/usbhid.ko
+kernel/drivers/idle/i7300_idle.ko
kernel/drivers/input/input-core.ko
kernel/drivers/input/ff-memless.ko
kernel/drivers/input/mousedev.ko
@@ -173,21 +170,16 @@
kernel/drivers/input/serio/serio_raw.ko
kernel/drivers/leds/led-class.ko
kernel/drivers/macintosh/mac_hid.ko
-kernel/drivers/mfd/mfd-core.ko
kernel/drivers/misc/tifm_core.ko
kernel/drivers/misc/tifm_7xx1.ko
kernel/drivers/misc/cb710/cb710.ko
kernel/drivers/mmc/card/mmc_block.ko
-kernel/drivers/mmc/card/mmc_test.ko
kernel/drivers/mmc/card/sdio_uart.ko
kernel/drivers/mmc/core/mmc_core.ko
kernel/drivers/mmc/host/sdhci.ko
kernel/drivers/mmc/host/sdhci-pci.ko
kernel/drivers/mmc/host/sdhci-pci-data.ko
kernel/drivers/mmc/host/sdhci-pci-o2micro.ko
-kernel/drivers/mmc/host/sdhci-acpi.ko
-kernel/drivers/mmc/host/sdhci-pxav3.ko
-kernel/drivers/mmc/host/sdhci-pxav2.ko
kernel/drivers/mmc/host/wbsd.ko
kernel/drivers/mmc/host/tifm_sd.ko
kernel/drivers/mmc/host/cb710-mmc.ko
@@ -209,6 +201,8 @@
kernel/drivers/net/ethernet/dec/tulip/winbond-840.ko
kernel/drivers/net/ethernet/dec/tulip/de4x5.ko
kernel/drivers/net/ethernet/dlink/sundance.ko
+kernel/drivers/net/ethernet/intel/e100.ko
+kernel/drivers/net/ethernet/intel/e1000e/e1000e.ko
kernel/drivers/net/ethernet/intel/igbvf/igbvf.ko
kernel/drivers/net/ethernet/natsemi/natsemi.ko
kernel/drivers/net/ethernet/natsemi/ns83820.ko
@@ -227,6 +221,8 @@
kernel/drivers/net/ppp/pppox.ko
kernel/drivers/net/ppp/pppoe.ko
kernel/drivers/net/slip/slhc.ko
+kernel/drivers/pps/pps_core.ko
+kernel/drivers/ptp/ptp.ko
kernel/drivers/rtc/rtc-cmos.ko
kernel/drivers/scsi/scsi_mod.ko
kernel/drivers/scsi/sd_mod.ko
@@ -254,8 +250,6 @@
kernel/drivers/video/fbdev/core/cfbfillrect.ko
kernel/drivers/video/fbdev/core/cfbcopyarea.ko
kernel/drivers/video/fbdev/c
la parola agli esperti.
Grazie!
--------------------------------------------------------------------------
Hi, doing some testing on a tablet with intel baytrail processor, I realized that the 64bit version does not detect memory (mmc).
I would like to have both working if it is possible.
Comparing the built-in modules of the two kernels I get this
see above
the word to the experts.
Thank you!
Offline
Hi Gibor,
If understand, 32bit version work but not 64bit ?
Maybe related to this mods, but not sure :
[c]
-kernel/drivers/mmc/host/sdhci-acpi.ko
-kernel/drivers/mmc/host/sdhci-pxav3.ko
-kernel/drivers/mmc/host/sdhci-pxav2.ko
[/c]
For my current kernel 4.19.298, config for mmc :
[c]
# CONFIG_PCI_GOMMCONFIG is not set
CONFIG_PCI_MMCONFIG=y
# Supported MMC/SDIO adapters
CONFIG_MMC=y
CONFIG_MMC_BLOCK=y
CONFIG_MMC_BLOCK_MINORS=8
# CONFIG_MMC_TEST is not set
# MMC/SD/SDIO Host Controller Drivers
# CONFIG_MMC_DEBUG is not set
CONFIG_MMC_SDHCI=y
CONFIG_MMC_SDHCI_PCI=y
# CONFIG_MMC_RICOH_MMC is not set
# CONFIG_MMC_SDHCI_ACPI is not set
CONFIG_MMC_SDHCI_PLTFM=y
# CONFIG_MMC_SDHCI_F_SDH30 is not set
CONFIG_MMC_WBSD=y
CONFIG_MMC_TIFM_SD=y
CONFIG_MMC_SDRICOH_CS=m
CONFIG_MMC_CB710=y
CONFIG_MMC_VIA_SDMMC=y
CONFIG_MMC_VUB300=y
CONFIG_MMC_USHC=y
CONFIG_MMC_USDHI6ROL0=y
CONFIG_MMC_CQHCI=y
# CONFIG_MMC_TOSHIBA_PCI is not set
# CONFIG_MMC_MTK is not set
# CONFIG_MMC_SDHCI_XENON is not set
# CONFIG_MMC35240 is not set
[/c]
Offline
Ciao Shann, non è proprio così. Le versioni ufficiali non riconoscono le memorie mmc su quel tablet. Quella utilizzata da me a 32bit è un cavallo di battaglia utilizzato anche per i backup, e quel kernel l'ho ricompilato parecchie volte per varie esigenze. Il problema è che ho forti difficoltà a fare la stessa cosa con la versione a 64bit. Debbo usare uclibc su di altro computer, oppure installazioni a 64bit hanno gcc recenti che procurano svariati errori di tipo multiple definition ecc.
Con il post chiedevo se fosse possibile integrare nel kernel ufficiale quei moduli necessari, senza arrecare danno per tutti gli altri.
Ti allego l'output delle versioni sullo stesso tablet.
Slitaz-rolling-core official
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINTS
sda 8:0 1 1.8G 0 disk
└─sda1 8:1 1 1.8G 0 part
zram0 251:0 0 116.2M 0 disk [SWAP]
zram1 251:1 0 116.2M 0 disk [SWAP]
zram2 251:2 0 116.2M 0 disk [SWAP]
zram3 251:3 0 116.2M 0 disk [SWAP]
SliTaz GNU/Linux - 5.0-RC4 20231119
Linux slitaz 3.16.55-slitaz #4 SMP Thu Jul 14 18:31:26 UTC 2022 i686 GNU/Linux
slitaz-rolling-core64 official
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINTS
sda 8:0 1 1.8G 0 disk
└─sda1 8:1 1 1.8G 0 part
zram0 251:0 0 111.4M 0 disk [SWAP]
zram1 251:1 0 111.4M 0 disk [SWAP]
zram2 251:2 0 111.4M 0 disk [SWAP]
zram3 251:3 0 111.4M 0 disk [SWAP]
SliTaz GNU/Linux - 5.0-RC4 20231119
Linux slitaz 3.16.55-slitaz64 #2 SMP Thu Jul 14 18:20:24 UTC 2022 x86_64 GNU/Linux
slitaz-rolling-backup (rolling-core at 32bit mod)
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
sda 8:0 1 1,9G 0 disk
└─sda1 8:1 1 1,9G 0 part
mmcblk0 179:0 0 14,7G 0 disk
├─mmcblk0p1 179:1 0 300M 0 part
└─mmcblk0p2 179:2 0 14,4G 0 part
mmcblk0boot0 179:8 0 4M 1 disk
mmcblk0boot1 179:16 0 4M 1 disk
mmcblk0rpmb 179:24 0 512K 0 disk
mmcblk1 179:32 0 29,1G 0 disk
└─mmcblk1p1 179:33 0 29,1G 0 part
SliTaz GNU/Linux - 5.0-RC4 20220206
Linux slitaz 3.16.55-slitaz #5 SMP Sat Nov 25 17:51:01 CET 2023 i686 GNU/Linux
Hi Shann, that's not really the case. The official versions do not recognize mmc memories on that tablet. The one I use in 32bit is a workhorse also used for backups, and that kernel I have recompiled several times for various needs. The problem is that I have great difficulty doing the same thing with the 64bit version. I have to use uclibc on of other computer, or 64bit installations have recent gcc which give various errors like multiple definition etc.
With the post I was asking if it was possible to integrate those necessary modules into the official kernel, without causing harm to everyone else.
I attach the output of the versions on the same tablet.
Offline
Questa patch sembra bypassare il problema di gcc 11.2.0
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=748577
ora è in compilazione, vedremo quando finirà...
Offline
Hi gibor,
For sure, i think we can add necessary stuff for mmc on kernel config as module.
Normally we have already linux-mmc package.
With your test, it's seem few drivers miss when build
Offline
Hi Gibor,
I build 3.16.55-slitaz64 (build with gcc 6.3.0) with linux-slitaz.config64 + mmc as :
https://people.slitaz.org/~shann/linux64-mmc/linux64-3.16.55_mmc.tar
diff from repo config (both 32 / 64) :
[c]
root@slitaz:/home/slitaz/src/linux-3.16.55# diff -Bb linux-slitaz.config .config |grep MMC
-# CONFIG_PCI_GOMMCONFIG is not set
CONFIG_PCI_MMCONFIG=y
# Supported MMC/SDIO adapters
CONFIG_MMC_SDHCI=y
CONFIG_MMC_SDHCI_PCI=y
-# CONFIG_MMC_RICOH_MMC is not set
-# CONFIG_MMC_SDHCI_ACPI is not set
+CONFIG_MMC_RICOH_MMC=y
+CONFIG_MMC_SDHCI_ACPI=m
CONFIG_MMC_SDHCI_PLTFM=y
-# CONFIG_MMC_SDHCI_PXAV3 is not set
-# CONFIG_MMC_SDHCI_PXAV2 is not set
+CONFIG_MMC_SDHCI_PXAV3=m
+CONFIG_MMC_SDHCI_PXAV2=m
CONFIG_MMC_WBSD=y
CONFIG_MMC_TIFM_SD=y
CONFIG_MMC_SDRICOH_CS=m
[/c]
Tell me if it's ok or not for mmc, and adjust config on repo to add them.
Offline
Ciao Shann, grazie delle prove, ma ho avuto dei contrattempi con quello stupido EFI e la mia compilazione. Appena ho dato il make install anziché limitarsi a copiare il vmlinuz nella cartella del boot, ha alterato il grub, l'initrd, e l'efi. Dopo vari tentativi senza successo ho dovuto ripristinare un'immagine di originale del sistema.
Appena posso proverò il tuo kernel modificato su di una live rolling64.
È vero che i moduli ci sono, ma è anche vero che l'efi (stupido) avvia solo da una partizione in fat32, poi da lì il kernel deve già avere il necessario per riconoscere la root sulla mmcblk per poterla montare con l'init, e trovare cosi i moduli. Se non riesce a fare questo si blocca in kernel panic...
Hi Shann, thanks for the evidence, but I had some mishaps with the stupid EFI and my compilation. As soon as I gave the make install instead of just copying the vmlinuz to the boot folder, it altered the grub, initrd, and efi. After several unsuccessful attempts I had to restore an image of original system.
As soon as I can I will try your modified kernel on a live rolling64.
It is true that the modules are there, but it is also true that efi (stupid) only boots from a fat32 partition, then from there the kernel must already have the necessary to recognize the root on the mmcblk in order to mount it with init, and thus find the modules. If it can't do this it crashes in kernel panic....
Offline
Ciao Shann, dopo un po' di tempo riassumo le prove: Il kernel che hai postato non rileva le mmc.
Con uclibc sono riuscito a ricompilare il 3.16.55 a 64bit; questo a volte rileva la memoria, a volte no. Inserendo rootdelay=6 migliora, ma confonde spesso le due memorie fallendo l'avvio (microsd e emmc, entrambe rilevate come mmcblkxpx).
L'utilizzo dell'initrd assicura il funzionamento rilevandole con UUID, ma perdo i vantaggi dell'avvio diretto come di consueto fa slitaz.
Concludendo manterrò il kernel di slitaz per la live dei backup, mentre per l'installazione continuerò con il kernel fornito dalla concorrenza (*buntu).
In ogni caso la meraviglia è che con 3W ho un pc a 64bit quadcore completo. Impensabile finora con notebook o netbook da me posseduti (potenza minima 30W con un celeron a 32bit).
Buone feste a tutti!
Hi Shann, after some time I summarize the evidence: The kernel you posted does not detect mmc.
With uclibc I was able to recompile 3.16.55 to 64bit; this sometimes detects the memory, sometimes not. Inserting rootdelay=6 improves, but it often confuses the two memories by failing to boot (microsd and emmc, both detected as mmcblkxpx).
Using initrd ensures operation by detecting them with UUID, but I lose the benefits of direct booting as slitaz usually does.
In conclusion I will keep the slitaz kernel for live backups, while for installation I will continue with the kernel provided by the competition (*buntu).
In any case the wonder is that with 3W I have a full quadcore 64bit pc. Unthinkable so far with notebooks or netbooks I have owned (minimum power 30W with a 32bit celeron).
Happy holidays to all!
Offline
[ Generated in 0.022 seconds, 7 queries executed - Memory usage: 1.57 MiB (Peak: 1.77 MiB) ]
