CT-PC89E Bugs - Revision history

Peter Huewe: Added to category CT-PC89E

2011-04-13T21:47:42Z

Added to category CT-PC89E

Lkcl: /* Bugs found with operating the CT-PC89E */

2010-03-21T19:22:54Z

‎Bugs found with operating the CT-PC89E

Lkcl: /* Bugs found with operating the CT-PC89E */

2010-03-19T19:08:21Z

‎Bugs found with operating the CT-PC89E

Lkcl: /* Bugs found with operating the CT-PC89E */

2010-03-12T13:21:40Z

‎Bugs found with operating the CT-PC89E

Lkcl: /* 9mar2010 report from bjdooks = */

2010-03-09T21:05:40Z

‎9mar2010 report from bjdooks =

Lkcl: /* 21mar2010 report from fjp */

2010-03-09T21:01:06Z

‎21mar2010 report from fjp

Lkcl: /* Bugs found with operating the CT-PC89E */

2010-03-04T16:58:04Z

‎Bugs found with operating the CT-PC89E

Lkcl: /* Bugs found with operating the CT-PC89E */

2010-03-03T12:46:10Z

‎Bugs found with operating the CT-PC89E

Lkcl: /* Bugs found with operating the CT-PC89E */

2010-03-03T12:44:39Z

‎Bugs found with operating the CT-PC89E

Lkcl: /* 21mar2010 report from fjp */

2010-03-02T22:17:50Z

‎21mar2010 report from fjp

← Older revision		Revision as of 21:47, 13 April 2011
Line 56:		Line 56:

	* I also have had multiple times that kbd and touchpad are dead after boot.		* I also have had multiple times that kbd and touchpad are dead after boot.
		+
		+
		+
		+
		+	[[Category: CT-PC89E]]

@@ Line 3: / Line 3: @@
 This page contains a record of problems found so far with the CT-PC89E, a very new design of ARM 8.9in netbook.  As they are corrected, the issues will be removed.
 * Starting "appliaction" spelling mistake
 * Plastic button activiating Reset pushed in (!) after use only about ten times.  still works but care now needed.
 * No lid "sleep" system, so LCD stays active after lid is closed
 * layout of filesystem is not FHS compliant.
 * kernel modules have been moved to /drivers instead of /usr/lib/modules/2.6.24.2/kernel/drivers, violating FHS compliance and absolutely guaranteeing future problems with system upgrades.

@@ Line 25: / Line 25: @@
 * lkcl: after dropping the machine (entirely my own fault!) onto a wooden floor from a height of 15in, it fell onto the power lead and ethernet dongle side.  the ethernet dongle is now also fairly loose, and requires to have gravity-assistance (dangling) in order to maintain an ethernet connection.  not entirely sure what to recommend, here, because the connector is about as small as you can get, and there's simply no room for the height or width of a "standard" RJ45 connector, because overall the machine is really very thin.  the macbook "air" design springs to mind, with a hole in the underside and an approx 30 degree angle on the RJ45 connector, but... on a low-cost netbook, that would seem to be a bit of a mad design decision.
 * the NAND flash is compiled with ext4 tools, then marked as ext3.  the kernel is then only compiled with ext4 driver, from 2.6.24.2 which is KNOWN to be UNSTABLE.  this will (and is already) cause problems.
 == 21feb2010 report from fjp ==

@@ Line 24: / Line 24: @@
 * removing ethernet cable, re-insertion doesn't result in reconnection.  this may be due to linux kernel bugs, but it is hard to tell
 * lkcl: after dropping the machine (entirely my own fault!) onto a wooden floor from a height of 15in, it fell onto the power lead and ethernet dongle side.  the ethernet dongle is now also fairly loose, and requires to have gravity-assistance (dangling) in order to maintain an ethernet connection.  not entirely sure what to recommend, here, because the connector is about as small as you can get, and there's simply no room for the height or width of a "standard" RJ45 connector, because overall the machine is really very thin.  the macbook "air" design springs to mind, with a hole in the underside and an approx 30 degree angle on the RJ45 connector, but... on a low-cost netbook, that would seem to be a bit of a mad design decision.
 == 21feb2010 report from fjp ==

@@ Line 42: / Line 42: @@
 * evbug is compiled in causing major spam of dmesg (making it unusable). It is also a security issue as it means all keystrokes get logged. And it probably reduces performance somewhat.
-== 9mar2010 report from bjdooks ===
+== 9mar2010 report from bjdooks ==
 related to the DM9000, from the photos of the PCB.  ben has done a design: http://www.fluff.org/ben/pics/im-s3e-mk1/img_0472.jpg

@@ Line 22: / Line 22: @@
 * the built-in microphone doesn't seem to work, or if it does, the kernel drivers are borked and don't make it available under ACPI.  speakers work fine though.
 * The newly-developed http://gitorious.org/xf86-video-s3c64xx which is 20x or greater speedups cannot be used because the linux kernel source is not yet available, in order to add it.
-* removing ethernet cable, re-insertion doesn't result in reconnection
+* removing ethernet cable, re-insertion doesn't result in reconnection.  this may be due to linux kernel bugs, but it is hard to tell
 == 21mar2010 report from fjp ==

@@ Line 8: / Line 8: @@
 * machine left on overnight resulted in "LCD burn" probably due to kernel crash resulted in brightness of LCD being pushed beyond limits (a common problem on embedded systems).  can only really be fixed by a hardware watchdog timer, resulting in power-down.
 * layout of filesystem is not FHS compliant.
-* kernel modules have been moved to /drivers instead of /usr/lib/modules/2.6.24.2/kernel/drivers
+* kernel modules have been moved to /drivers instead of /usr/lib/modules/2.6.24.2/kernel/drivers, violating FHS compliance and absolutely guaranteeing future problems with system upgrades.
 * debian packages on MOS mid-fun operating system violate FHS and standard debian policy: many packages install identical files TWICE, some of which are in an indentical file-structure under /READONLY_FS.  this should NOT be done: the files should be copied manually, or installed using debootstrap into /READONLY_FS but NOT added twice into one package.
 * kernel modules compiled into the kernel are still present in /drivers, causing segfaults in some cases

@@ Line 9: / Line 9: @@
 * layout of filesystem is not FHS compliant.
 * kernel modules have been moved to /drivers instead of /usr/lib/modules/2.6.24.2/kernel/drivers
-* debian packages on MOS mid-fun operating system violate FHS and standard debian policy: many packages install identical files TWICE, some of which are in an indentical file-structure under /READONLY_FS.
+* debian packages on MOS mid-fun operating system violate FHS and standard debian policy: many packages install identical files TWICE, some of which are in an indentical file-structure under /READONLY_FS.  this should NOT be done: the files should be copied manually, or installed using debootstrap into /READONLY_FS but NOT added twice into one package.
 * kernel modules compiled into the kernel are still present in /drivers, causing segfaults in some cases
 * kernel GPL source code not yet been provided
 * u-boot GPL source code not yet been provided
-* MOS mid-fun GPL source code not yet been provided (workaround provided by creating a Debian/Lenny image)
+* MOS mid-fun GPL source code not yet been provided (workaround provided by deleting MOS entirely and creating a Debian/Lenny image instead)
-* on 3G dialup (vodafone USB modem), pppd segfaults (presumably kernel problem) under Debian, and machine becomes unreliable (indicating kernel problem).
+* on 3G dialup (vodafone USB modem), pppd segfaults (presumably kernel problem) under Debian, and machine becomes unreliable (indicating kernel problem).  this can be fixed by providing the kernel source code.
 * on 3G dialup (USB modem used by adam), again it fails.  kernel source DEFINITELY needed.
 * root password was cracked within 4 days, by running "john" on /etc/passwd, and found to be "mos2010"
 * /etc and many other files are owned by midfun not root, making it unnecessary to have performed the crack: just edit /etc/passwd
 * /sbin/udevd and other applications run as root are owned by midfun, allowing them to be replaced and perform root operations under user control.
-* screen brightness control is non-linux-compliant: there is no standard entry (that could be found) in /proc or /sys for the LCD brightness, making it impossible to use standard linux applications and distributions.
+* screen brightness control is non-linux-compliant: there is no standard entry (that could be found) in /proc or /sys for the LCD brightness, making it impossible to use standard linux applications and distributions.  this can be fixed by providing the linux kernel source code.
 * the built-in microphone doesn't seem to work, or if it does, the kernel drivers are borked and don't make it available under ACPI.  speakers work fine though.
-* The newly-developed http://gitorious.org/xf86-video-s3c64xx which is 20x or greater speedups cannot be used because the linux kernel source is not yet available.
+* The newly-developed http://gitorious.org/xf86-video-s3c64xx which is 20x or greater speedups cannot be used because the linux kernel source is not yet available, in order to add it.
 * removing ethernet cable, re-insertion doesn't result in reconnection

← Older revision		Revision as of 22:17, 2 March 2010
Line 39:		Line 39:
	** - PowerManager Eanble		** - PowerManager Eanble
	** - FALSH (instead of FLASH)		** - FALSH (instead of FLASH)
		+	* evbug is compiled in causing major spam of dmesg (making it unusable). It is also a security issue as it means all keystrokes get logged. And it probably reduces performance somewhat.