4 ms·

I'll believe that real progress has been made when google updates the major kernel version for a phone. No OEM (AFAIK) has ever done a kernel (major version) bu

I'll believe that real progress has been made when google updates the major kernel version for a phone. No OEM (AFAIK) has ever done a kernel (major version) bump.

culot 8y ago
It was either the original Moto G or the 2nd gen one that got a kernel version bump - from 3.4 to 3.10, I think. I remember being impressed at the time. Newer Sony devices got a major version bump from 4.4 to 4.9 with their Pie releases: https://developer.sony.com/develop/open-devices/latest-updates https://developer.sony.com/develop/open-devices/latest-updat... I wonder if the major version update for Sony's 9.0/Pie was due to a Qualcomm BSP change, or if it was solely their initiative? I think the reason for typically not doing major kernel updates may be down to sticking with reference stuff from BSPs. (I don't think they're offering any big updates, not kernel or Android major versions, for their Mediatek stuff. I think that might be typical though.) Kind of cool, anyway, to move from one LTS kernel release to the next. I wonder have any other OEMs made the same move?
corbet 8y ago
The Galaxy S and Galaxy Nexus both got major version bumps. Old history at this point, but it does show that this can happen... (https://lwn.net/Articles/738225/ https://lwn.net/Articles/738225/)
fffrantz 8y ago
Honestly, it's extremely difficult to update the kernel since it's usually tied to the BSP the SoC vendor is providing. Also, you gotta keep in mind the difficulties associated with the lazy 3rd parties providing drivers for a single kernel version, making updates almost impossible. -- Disclosure : I work for an OEM as a system architect, and it's been one of my goals to try and get newer kernel versions running. (Suffice to say I failed so far)
- baybal2 8y ago
  Doing so will require a complete shift of responsibilities for parties involved. I myself witnessed many "brains being broken" scenes when somebody tries to explain why Linux kernel development being done for "free", to men with educational background in business. In the end, you will never be successful doing so, be sure 99%. But whether they understood the bigger picture of things or not, technical needs need to supersede business needs, or that tech business will not work.
- pedrocr 8y ago
  Is upstreaming hardware drivers to at least staging quality that much of an uphill task? Seems like this should be fixed at the source by having the initial development go straight into mainline. That's what seems to happen for PC hardware anyway.
  - de_watcher 8y ago
    Upstreaming costs like several times more than just hack it and ship it. It seems that a lot of vendors can't estimate the costs and benefits.
    - fffrantz 8y ago
      That's exactly it. And even keeping to the preferred/qualified vendor list provided by the SoC vendor does not guarantee that the drivers will be in the mainline... Android's use of the kernel really is fundamentally broken
izacus 8y ago
The article actually addresses that - why would Google (or Samsung or anyone really) risk device issues by doing this kind of upgrade? What is there to gain on a running device except angry users due to breakages of already working functionality? There's an upwards trend of people DISABLING updates because they tend to break tools they use in daily lives.
- bubblethink 8y ago
  >What is there to gain on a running device except angry users due to breakages of already working functionality? That's true, but regular distros are also faced with the same dilemma, and you see both cases. RH will patch a kernel to death ship of Theseus style for a decade, whereas Canonical offers newer stable kernel releases (although they also maintain the original kernel that the distro shipped with). Canonical leans more towards updating to a newer kernel version though. >There's an upwards trend of people DISABLING updates because they tend to break tools they use in daily lives. The general public is affected more by questionable updates higher up in the stack. I doubt if they interact directly with the kernel.
  - izacus 8y ago
    > The general public is affected more by questionable updates higher up in the stack. I doubt if they interact directly with the kernel. Things like bluetooth, OpenGL drivers and Wifi drivers are usually the pain points which love to break (or more accurately: break in a different way) when upgrading them. This causes those endless "I've updated my iPhone/Android phone and now it doesn't work with my car anymore!" posts. It's usually preferrable to keep the things working as they are, since we humans accept that more easily.

I'll believe that real progress has been made when google updates the major kernel version for a phone. No OEM (AFAIK) has ever done a kernel (major version) bump.

culot 8y ago

It was either the original Moto G or the 2nd gen one that got a kernel version bump - from 3.4 to 3.10, I think. I remember being impressed at the time. Newer Sony devices got a major version bump from 4.4 to 4.9 with their Pie releases: https://developer.sony.com/develop/open-devices/latest-updates https://developer.sony.com/develop/open-devices/latest-updat... I wonder if the major version update for Sony's 9.0/Pie was due to a Qualcomm BSP change, or if it was solely their initiative? I think the reason for typically not doing major kernel updates may be down to sticking with reference stuff from BSPs. (I don't think they're offering any big updates, not kernel or Android major versions, for their Mediatek stuff. I think that might be typical though.) Kind of cool, anyway, to move from one LTS kernel release to the next. I wonder have any other OEMs made the same move?

corbet 8y ago

The Galaxy S and Galaxy Nexus both got major version bumps. Old history at this point, but it does show that this can happen... (https://lwn.net/Articles/738225/ https://lwn.net/Articles/738225/)

fffrantz 8y ago

Honestly, it's extremely difficult to update the kernel since it's usually tied to the BSP the SoC vendor is providing. Also, you gotta keep in mind the difficulties associated with the lazy 3rd parties providing drivers for a single kernel version, making updates almost impossible. -- Disclosure : I work for an OEM as a system architect, and it's been one of my goals to try and get newer kernel versions running. (Suffice to say I failed so far)

baybal2 8y ago
Doing so will require a complete shift of responsibilities for parties involved. I myself witnessed many "brains being broken" scenes when somebody tries to explain why Linux kernel development being done for "free", to men with educational background in business. In the end, you will never be successful doing so, be sure 99%. But whether they understood the bigger picture of things or not, technical needs need to supersede business needs, or that tech business will not work.
pedrocr 8y ago
Is upstreaming hardware drivers to at least staging quality that much of an uphill task? Seems like this should be fixed at the source by having the initial development go straight into mainline. That's what seems to happen for PC hardware anyway.
- de_watcher 8y ago
  Upstreaming costs like several times more than just hack it and ship it. It seems that a lot of vendors can't estimate the costs and benefits.
  - fffrantz 8y ago
    That's exactly it. And even keeping to the preferred/qualified vendor list provided by the SoC vendor does not guarantee that the drivers will be in the mainline... Android's use of the kernel really is fundamentally broken

izacus 8y ago

The article actually addresses that - why would Google (or Samsung or anyone really) risk device issues by doing this kind of upgrade? What is there to gain on a running device except angry users due to breakages of already working functionality? There's an upwards trend of people DISABLING updates because they tend to break tools they use in daily lives.

bubblethink 8y ago
>What is there to gain on a running device except angry users due to breakages of already working functionality? That's true, but regular distros are also faced with the same dilemma, and you see both cases. RH will patch a kernel to death ship of Theseus style for a decade, whereas Canonical offers newer stable kernel releases (although they also maintain the original kernel that the distro shipped with). Canonical leans more towards updating to a newer kernel version though. >There's an upwards trend of people DISABLING updates because they tend to break tools they use in daily lives. The general public is affected more by questionable updates higher up in the stack. I doubt if they interact directly with the kernel.
- izacus 8y ago
  > The general public is affected more by questionable updates higher up in the stack. I doubt if they interact directly with the kernel. Things like bluetooth, OpenGL drivers and Wifi drivers are usually the pain points which love to break (or more accurately: break in a different way) when upgrading them. This causes those endless "I've updated my iPhone/Android phone and now it doesn't work with my car anymore!" posts. It's usually preferrable to keep the things working as they are, since we humans accept that more easily.