eMMC sudden death research

[Product F(RED)]

What does this have anything to do with Exynos/Qualcomm?

Sorry, I thought I was in the main SDS thread (I'm subscribed to this and both).

But while we're on this topic, to answer your question, the Qualcomm models have only had a few instances of SDS. And I live in the US, so warranty repair or replacement is easier for me.

Again, sorry. I posted in the wrong thread.

 

[AndreiLux]

I think he thinked qualcomm is much error free than that **** from samsung

Sent from my GT-I9300 using xda premium

Sorry, I thought I was in the main SDS thread (I'm subscribed to this and both).

But while we're on this topic, to answer your question, the Qualcomm models have only had a few instances of SDS. And I live in the US, so warranty repair or replacement is easier for me.

Again, sorry. I posted in the wrong thread.

I wasn't aware that Qualcomm made eMMC modules! Learning something every day.

 

[Product F(RED)]

I wasn't aware that Qualcomm made eMMC modules! Learning something every day.

I believe the North American models use different eMMC chips than the ones that are in the SDS-prone batch of i9300's, but I don't think they're necessarily made by Qualcomm.

 

[Entropy512]

Sorry, I thought I was in the main SDS thread (I'm subscribed to this and both).

But while we're on this topic, to answer your question, the Qualcomm models have only had a few instances of SDS. And I live in the US, so warranty repair or replacement is easier for me.

Again, sorry. I posted in the wrong thread.

Pure coincidence, and not really related to this thread. There's nothing preventing a defective chip from showing up in a Qcom-based Samsung (see Hercules/Skyrocket and Superbrick).

 

[Product F(RED)]

Pure coincidence, and not really related to this thread. There's nothing preventing a defective chip from showing up in a Qcom-based Samsung (see Hercules/Skyrocket and Superbrick).

Oh, of course. I agree fully. I'm just saying that for the S3 specifically, SDS is far less frequent because of the difference in hardware (and most likely eMMC). It has been reported a few times, but it's nowhere near as ubiquitous as with the i9300/i9305.

 

[kaidokert]

So, i have suddenly became responsible for reviving my spouses SDSed GT-I9300 or at least pull data off the eMMC ( phone bought in US through Amazon so warranty is a predicament ). I have a day job, several hobbies and generally nice outlook in life and no interest hacking this, but "because my girlfriend's a vegitarian which pretty much makes me a vegitarian" situation here i am obliged.
As a last resort, i'm looking at desoldering the eMMC chip and wiring it into one of 3.3v dev boards that i have ( a bunch of cortex-M0s, Arduino Due, Pandaboard etc ) and from what i gather eMMC only requires 2 lines in minimum conf, CLK and DAT0 to talk to it so i could do abysmally slow raw data dump if i wanted to. But i only started reading up on the entire issue and glossed over eMMC spec for the first time 15 mins ago.
Any better pointers before i go down that last resort path ? The phone still goes to Odin mode, can i load a minimal kernel or u-boot or something through that somehow ?

 

[Entropy512]

So, i have suddenly became responsible for reviving my spouses SDSed GT-I9300 or at least pull data off the eMMC ( phone bought in US through Amazon so warranty is a predicament ). I have a day job, several hobbies and generally nice outlook in life and no interest hacking this, but "because my girlfriend's a vegitarian which pretty much makes me a vegitarian" situation here i am obliged.
As a last resort, i'm looking at desoldering the eMMC chip and wiring it into one of 3.3v dev boards that i have ( a bunch of cortex-M0s, Arduino Due, Pandaboard etc ) and from what i gather eMMC only requires 2 lines in minimum conf, CLK and DAT0 to talk to it so i could do abysmally slow raw data dump if i wanted to. But i only started reading up on the entire issue and glossed over eMMC spec for the first time 15 mins ago.
Any better pointers before i go down that last resort path ? The phone still goes to Odin mode, can i load a minimal kernel or u-boot or something through that somehow ?

The problem is that most likely, the eMMC is so hosed at a low level that the data is inaccesible.

There are some special tricks available (search around) for booting the I9300 off of a special SD card to repopulate the bootloader. So if you can replace the eMMC chip, you should be able to repopulate the bootloaders.

Another possibility is that if you can boot from an external SD, you MIGHT be able to send commands to resize the boot partitions downwards and back up. This will, in theory, cause a full chip reset. All data will be lost, but the damaged areas will get cleaned up. However, this is only theory based on reading the leaked datasheet - no one has EVER done this, so no one knows if it will work.

Ideally for these repairs we'd have an expert in JTAG recovery help out - I'm sure with the knowledge from that datasheet and a few dead devices, the RIFFbox guys could figure something out. I once emailed Josh at mobiletechvideos these new discoveries, but never heard back from him.

 

[Product F(RED)]

The problem is that most likely, the eMMC is so hosed at a low level that the data is inaccesible.

There are some special tricks available (search around) for booting the I9300 off of a special SD card to repopulate the bootloader. So if you can replace the eMMC chip, you should be able to repopulate the bootloaders.

Another possibility is that if you can boot from an external SD, you MIGHT be able to send commands to resize the boot partitions downwards and back up. This will, in theory, cause a full chip reset. All data will be lost, but the damaged areas will get cleaned up. However, this is only theory based on reading the leaked datasheet - no one has EVER done this, so no one knows if it will work.

Ideally for these repairs we'd have an expert in JTAG recovery help out - I'm sure with the knowledge from that datasheet and a few dead devices, the RIFFbox guys could figure something out. I once emailed Josh at mobiletechvideos these new discoveries, but never heard back from him.

It's been done on the original Note actually. Can't find the original thread but I remember reading it. The guy basically moved the partitions over.

 

[Entropy512]

It's been done on the original Note actually. Can't find the original thread but I remember reading it. The guy basically moved the partitions over.

"It" being what? - replacing the eMMC (yeah, 1-2 people have soldered new ones on, they needed JTAG to repop the bootloaders since there is no known way to repopulate bootloaders via USB or SD on 4210 devices - Many 4412 devices including I9300 can be recovered from hardbrick from SD) or fixing a damaged eMMC using boot partition resizes (i'm fairly certain no one has done this yet)

 

[Product F(RED)]

"It" being what? - replacing the eMMC (yeah, 1-2 people have soldered new ones on, they needed JTAG to repop the bootloaders since there is no known way to repopulate bootloaders via USB or SD on 4210 devices - Many 4412 devices including I9300 can be recovered from hardbrick from SD) or fixing a damaged eMMC using boot partition resizes (i'm fairly certain no one has done this yet)

No; like I said I saw a fix where the guy moved the partitions over so that the undamaged portions were unused.

 

[Entropy512]

No; like I said I saw a fix where the guy moved the partitions over so that the undamaged portions were unused.

That's a fugly hack that usually doesn't last long because the corrupted data structures seem to spread.

We're talking about ways to completely recover a device.

 

[AdamOutler]

@Oranav , @Rebellos and myself are working on a bootloader to work around this issue. https://github.com/Rebell/exynos4_uboot

We should be able to insert test commands to my bricked GS3, but I want to try to recover the data first. This device was given to me for data recovery and testing bootloaders. Currently we are working around the EMMC and booting from SDCard.

 

[vim1]

Hi , I am searching for peoples who reversed eMMC. We created HW solution to access eMMC directly in PCB and trying to revive bricked cards.

 

[hackerse7en]

is this the emmc ?
http://www.parts4repair.com/samsung-galaxy-note-ii-n7100-flash-chip-with-program/

if i buy this emmc and replace it in my n7100 will it work?

 

[Ricky Divjakovski]

@Oranav , @Rebellos and myself are working on a bootloader to work around this issue. https://github.com/Rebell/exynos4_uboot

We should be able to insert test commands to my bricked GS3, but I want to try to recover the data first. This device was given to me for data recovery and testing bootloaders. Currently we are working around the EMMC and booting from SDCard.

did you end up going through with the project? im curious?

 

[Rebellos]

We didn't finish it. Lack of time or something...

 

[n4naveen]

can someone tell me SDS bricked note 2 can be reversed or something with the existing eMMC ? ... I bought it in singapore which he claimed to have international warranty but back in india samsung service center guys reject to fix it under warranty.

 

[Entropy512]

can someone tell me SDS bricked note 2 can be reversed or something with the existing eMMC ? ... I bought it in singapore which he claimed to have international warranty but back in india samsung service center guys reject to fix it under warranty.

In theory, yes. In practice - no one has yet to put the required components together. (Actually, while the I9300 does have an SD-based "unbrick bootloader" process, I'm not sure if that works on Note2 units)

Pretty much two things are needed to recovery an SDSed or Superbricked device:
1) Use the boot partition resize trick to cause a low-level eMMC wipe - see Hashcode's success on the Kindle Fire - won't work if the device's eMMC is so corrupted it can't even boot a kernel or recovery
2) Once the eMMC is wiped, repopulate bootloaders - this is far harder to do on Exynos devices than on GP OMAPs like the Kindle Fire, which is why no one has done it yet - in theory this last part could be done by a JTAG service

 

[vim1]

In theory, yes. In practice - no one has yet to put the required components together. (Actually, while the I9300 does have an SD-based "unbrick bootloader" process, I'm not sure if that works on Note2 units)

Pretty much two things are needed to recovery an SDSed or Superbricked device:
1) Use the boot partition resize trick to cause a low-level eMMC wipe - see Hashcode's success on the Kindle Fire - won't work if the device's eMMC is so corrupted it can't even boot a kernel or recovery
2) Once the eMMC is wiped, repopulate bootloaders - this is far harder to do on Exynos devices than on GP OMAPs like the Kindle Fire, which is why no one has done it yet - in theory this last part could be done by a JTAG service

We ready to test your theory Please answer in AIM
We have done vendor boot size change with CMD62 on VYL00M

Command 62 (ARG: 0xEFAC62EC)
Command 62 (ARG: 0x00CBAEA7)
Command 62 (ARG: bla bla)

But none of "low-level wipe happened"

 

[Entropy512]

We ready to test your theory Please answer in AIM
We have done vendor boot size change with CMD62 on VYL00M

Command 62 (ARG: 0xEFAC62EC)
Command 62 (ARG: 0x00CBAEA7)
Command 62 (ARG: bla bla)

But none of "low-level wipe happened"

Interesting, it worked for the Kindle Fire guys - http://forum.xda-developers.com/showthread.php?t=2413453