Guest t0mm13b Posted January 13, 2011 Report Posted January 13, 2011 Please discuss: Would it be beneficial to unset a few things that caught my eye: CONFIG_DEBUG_INFO + CONFIG_DEBUG_KERNEL? (both in Kernel Hacking) - Would that see a speed gain? :D CONFIG_MAGIC_SYSRQ - (Kernel Hacking) Magic SysRq key - is that really needed? Like how would you do it on a touchscreen? :P CONFIG_DEBUG_BUGVERBOSE (Kernel Hacking) - That needed?CONFIG_DEBUG_USER + CONFIG_DEBUG_ERRORS (both in Kernel Hacking) - necessary?, maybe, kernel panics, can be logged to flash anyway by way of this CONFIG_APANIC (Device Drivers > Misc Devices) The way I see it is this - the last two is for if you're extremely hardcore kernel hacker, it would be handy alright, but on the face of it, we're already in an Android environment and those things would be pointless, so what if the handset is rooted or not, either way, for the average user, they're hardly going to see that... I, myself, like to tinker and have been pondering about those in the list above.... Any worthwhile thoughts/suggestions....?
Guest t0mm13b Posted January 13, 2011 Report Posted January 13, 2011 Some more after exploring the options: CONFIG_SCSI_LOGGING + SCSI_LOWLEVEL ( Device Drivers > SCSI Device Support > SCSI low-level Drivers)CONFIG_SMC91X (Device Drivers > Network device support > Ethernet (10 or 100Mbit) - Have no idea - that required?)CONFIG_OV5642 (Device Drivers > Multimedia Support > Video Capture adapters > Sensor ov5642 (YUV 5.0M)) - Assuming that's the camera 5Mp, not required for S.F. There's Radio Adapter - unchecked - presumably that could explain radio support not working, despite DAB set...?! Any other ideas?!
Guest Rotmann Posted January 13, 2011 Report Posted January 13, 2011 Nice thread, hopefully it will get the attention of the devs. Very interesting, keep up the good work! :D
Guest t0mm13b Posted January 13, 2011 Report Posted January 13, 2011 (edited) Nice thread, hopefully it will get the attention of the devs. Very interesting, keep up the good work! :P Thanks :D I have spotted a few things that's making my head spin.... Forgive the questioning of it...Clockwork recovery seems to overwrite the original kernel build's command line on boot, with this one, and its different - question is "Is Clockwork recovery boot up procedure hardwired to specify this - rdinit=/init console=ttyDCC0 console=ttyMSM2,115200", I am under the impression that using ttyMSM2,115200 will allow you to somehow see the kernel boot messages whizzing by? How does one do it, this would be a useful thing to have to determine if everything is hunky-dory... nonetheless, the compile type build of the kernel has this "CONFIG_CMDLINE="init=/sbin/init root=/dev/ram rw initrd=0x11000000,16M console=ttyDCC0 mem=88M"On another ROM - I think it was Customised BladeVillain (Eclair), there was a setting under Applications to not to charge the battery when plugged in, in Modaco's ROM that is not present - any chance of putting that back into it - this could be useful to be able to share files via cable without screwing up the battery by constantly plugging it in and defaulting to charging...I have integrated the driver for Atheros R6000 driver for the chipset so that it can be built, the github page for the kernel... The only thing about this is, "where and how did Paul get the binary firmware and patch for that Modaco rom - is it needed as the core driver is now available and can be recompiled?" there's a wifi directory under /system, that has athtcmd_ram.binathwlan.bin.z77data.patch.hw2_0.binWould they be needed? Hmmm.... I will have to find out and re-do up the Modaco ROM for further investigation.... I will post more findings on this... Edited January 13, 2011 by t0mm13b
Guest oh!dougal Posted January 13, 2011 Report Posted January 13, 2011 (edited) Two points -- 5mp cameras are in some European Blades (eg Finland) -- don't chuck anything just because it concerns the 5mp ! -- regarding TTY and boot messages - see http://android.modaco.com/content/zte-blad...2-on-the-blade/ AND (added) CONFIG_MAGIC_SYSRQ - (Kernel Hacking) Magic SysRq key - is that really needed? Like how would you do it on a touchscreen? blink.gif With host mode USB, you can connect a full usb keyboard ... And there's more mainstream ways to attach a Bluetooth one. Perhaps some of the 'left-over' debug facilities could be reopened to assist aftermarket devs? Edited January 13, 2011 by oh!dougal
Guest t0mm13b Posted January 13, 2011 Report Posted January 13, 2011 Two points -- 5mp cameras are in some European Blades (eg Finland) -- don't chuck anything just because it concerns the 5mp ! -- regarding TTY and boot messages - see http://android.modaco.com/content/zte-blad...2-on-the-blade/ AND (added) With host mode USB, you can connect a full usb keyboard ... And there's more mainstream ways to attach a Bluetooth one. Perhaps some of the 'left-over' debug facilities could be reopened to assist aftermarket devs? Cheers +1 for that... Thanks! :D
Guest Ippe H Posted January 14, 2011 Report Posted January 14, 2011 Is is possible to get wimax working on blade? blade-config # CONFIG_WIMAX is not set
Guest t0mm13b Posted January 14, 2011 Report Posted January 14, 2011 (edited) Right, have updated github on this and included AR6000 driver and ATH6KL drivers also... This is the configuration I used that runs against the github's kernel source in the attached... Fingers crossed now that I'll reboot it.... ;)buggy.config.gz Edited January 15, 2011 by t0mm13b
Guest t0mm13b Posted January 14, 2011 Report Posted January 14, 2011 Meh! Back to drawing board.... *sigh* Something got b0rk3d :) :D Will re-check it again... On boot up, the kernel "sort of" went into feedback loop - more likely panic'd and anyway most of debug logs etc was removed.... lulz!!! ;)
Guest MDCFan Posted January 14, 2011 Report Posted January 14, 2011 (edited) Is is possible to get wimax working on blade? blade-config # CONFIG_WIMAX is not set Blade does not have any WiMAX hardware. You can have WiMAX support in the kernel, but without the actual transceiver it is useless. Edited January 14, 2011 by MDCFan
Guest t0mm13b Posted January 14, 2011 Report Posted January 14, 2011 (edited) Whooooooo!!!!! Wheeeeeee!!!!! Tweaked Modaco Rom7 with kernel from the said github repo, here's the attached Boot up and running ok... will keep you posted.... ;)tweaked_modaco_rom7.config.gz Edited January 15, 2011 by t0mm13b
Guest t0mm13b Posted January 15, 2011 Report Posted January 15, 2011 Right, the kernel booted up successfully, so far so good.... Switch on Wifi... Whoops, tis b0rk3d... :) Nonetheless, the AR6000 driver is seriously out of whack and am mystified as to where Paul got the source from!! ;) Anyhoo... I have reverted back to original rom 8 baked from the kitchen.... and reported and attached my kernel tweak in the hope Paul takes note of it... Nonetheless, methinks its game over for me... but I hope that knowledge will be of help... If you're really curious to see the difference:Plug in the cable to your handsetrun adb shellfrom the adb shell prompt, issue zcat /proc/config.gz > /sdcard/current_configExit out of the adb shellIssue adb pull /sdcard/current_config on your current command lineFinally, run a diff on the configuration used with the one I've attached in the linky above in which you have to follow to get at the attached tweaked_config.gz You will see how much noise is generated by the kernel... am thinking, increase in performance, battery saving as those noises can cost a lot of overhead in terms of milliseconds, may not be noticeable to us mere mortals....
Guest oh!dougal Posted January 15, 2011 Report Posted January 15, 2011 (edited) Just as a side note I was wondering whether CONFIG_CPU_FREQ_DEFAULT_GOV_PERFORMANCE=y # CONFIG_CPU_FREQ_DEFAULT_GOV_POWERSAVE is not set # CONFIG_CPU_FREQ_DEFAULT_GOV_USERSPACE is not set # CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND is not set # CONFIG_CPU_FREQ_DEFAULT_GOV_CONSERVATIVE is not set CONFIG_CPU_FREQ_GOV_PERFORMANCE=y CONFIG_CPU_FREQ_GOV_POWERSAVE=y CONFIG_CPU_FREQ_GOV_USERSPACE=y CONFIG_CPU_FREQ_GOV_ONDEMAND=y CONFIG_CPU_FREQ_GOV_CONSERVATIVE=y # CONFIG_CPU_IDLE is not set CONFIG_CPU_FREQ_MSM=y was of any consequence to battery run-time concerns ... Edited January 15, 2011 by oh!dougal
Guest fonix232 Posted January 15, 2011 Report Posted January 15, 2011 Clockwork recovery seems to overwrite the original kernel build's command line on boot, with this one, and its different - question is "Is Clockwork recovery boot up procedure hardwired to specify this - rdinit=/init console=ttyDCC0 console=ttyMSM2,115200", I am under the impression that using ttyMSM2,115200 will allow you to somehow see the kernel boot messages whizzing by? How does one do it, this would be a useful thing to have to determine if everything is hunky-dory... nonetheless, the compile type build of the kernel has this "CONFIG_CMDLINE="init=/sbin/init root=/dev/ram rw initrd=0x11000000,16M console=ttyDCC0 mem=88M" I would be interested in this, someone got it working? ;)
Guest t0mm13b Posted January 15, 2011 Report Posted January 15, 2011 Just as a side note I was wondering whether was of any consequence to battery run-time concerns ... Good catch!!! +1... Hmmm..... ;)
Guest oh!dougal Posted January 15, 2011 Report Posted January 15, 2011 # # OTG and related infrastructure # CONFIG_USB_OTG_UTILS=y # CONFIG_USB_GPIO_VBUS is not set CONFIG_USB_MSM_OTG_72K=y # CONFIG_NOP_USB_XCEIV is not set CONFIG_MMC=y # CONFIG_MMC_DEBUG is not set # CONFIG_MMC_PERF_PROFILING is not set CONFIG_MMC_UNSAFE_RESUME=y # CONFIG_MMC_EMBEDDED_SDIO is not set CONFIG_MMC_PARANOID_SD_INIT=y Ummm. Does this look like our 2.2 should actually have USB OTG as standard ? ;) :)
Guest Quu Posted January 15, 2011 Report Posted January 15, 2011 Could you please add those HUGELONGCONF messages into code-box or something, it takes like 3 minutes to scroll down :<
Guest t0mm13b Posted January 15, 2011 Report Posted January 15, 2011 (edited) I would be interested in this, someone got it working? :) I did manage to get the boot up messages by doing a simple job ;) Have a look here at pastebin.com cat /proc/last_kmsg > /sdcard/kernel_boot.msgs Incidentally that boot up messages are from R8 Modaco Rom.... :D Edited January 15, 2011 by t0mm13b
Guest fonix232 Posted January 15, 2011 Report Posted January 15, 2011 I did manage to get the boot up messages by doing a simple job ;) cat /proc/last_kmsg > /sdcard/kernel_boot.msgs Incidentally that boot up messages are from R8 Modaco Rom.... :) But could these boot messages displayed at boot, instead of greendroid? Á la Android 0.1? Also, a suggestion: use pastebin.com for these snippets, they really make the page load very slow, and scrolling is a pain in the ass!
Guest t0mm13b Posted January 15, 2011 Report Posted January 15, 2011 But could these boot messages displayed at boot, instead of greendroid? Á la Android 0.1? Also, a suggestion: use pastebin.com for these snippets, they really make the page load very slow, and scrolling is a pain in the ass! Sorry fonix232, I have re-edited all the postings from moi in this thread and now... scrolling is eliminated ;) Thanks for the heads up! :) :D
Guest Meizirkki Posted January 15, 2011 Report Posted January 15, 2011 Just as a side note I was wondering whether was of any consequence to battery run-time concerns ... A good find! CPU_IDLE was not enabled?? If that's the case I'm surprised the battery lasts even this long..
Guest fonix232 Posted January 15, 2011 Report Posted January 15, 2011 A good find! CPU_IDLE was not enabled?? If that's the case I'm surprised the battery lasts even this long.. But, if we enable CPU_IDLE, then, maybe, we can find the longest lasting smartphone ;)
Guest t0mm13b Posted January 15, 2011 Report Posted January 15, 2011 Could you please add those HUGELONGCONF messages into code-box or something, it takes like 3 minutes to scroll down :< Have re-edited the posts for that Quu ;)
Guest t0mm13b Posted January 15, 2011 Report Posted January 15, 2011 But, if we enable CPU_IDLE, then, maybe, we can find the longest lasting smartphone ;) I am not 100%, but.... let me ask you this? Does the handset have ACPI, as the two are linked... as per the help on this one in the make menuconfig for CPU_IDLE "If you're using an ACPI-enabled platform, you should say Y here"
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now