FP 30.0.0.113 on Legacy PC with SSE CPU

[CBA2]

- Legacy IBM PC 300PL Pentium III with SSE (only) CPU

- Windows XP Pro patched as far as possible

- Firefox 45.9.0 ESR (last FX version for SSE CPUs)

FlashPlayer 29.0.0.171 is now installed and runs well on both IE and Firefox, albeit a bit slow (which is to be expected).  Trying to update to 30.0.0.113 generates an installer error early on during the process: "Adobe® Flash® Player Installer/Uninstaller 30.0 r0 has encountered a problem and needs to close.  We are sorry for the inconvenience."

Question: does FlashPlayer 30.0.0.113 require CPUs with SSE2 or higher?  If it does, fine .. but, I need to know this before spending time on troubleshooting.

PS: please, no comments on the old hardware and/or outdated software .. that's not the issue here!

[Robert Mc Dowell]

"please, no comments on the old hardware and/or outdated software"

dont' worry we are very open minded here, for any case

[jeromiec83223024]

Personally, I wouldn't spend a lot of time troubleshooting this, but nothing you've provided points to an issue with an unsupported instruction set. WinXP SP2 doesn't support modern cryptography, so it might be failing because it can't validate a digital signature or something, too.  The registry on WinXP is notorious for getting corrupted, and it might just be filesystem corruption or something.  I used to wipe my WinXP machines about once a year, because they'd just get squirrely after a while.  There's a lot of stuff that can go wrong that you don't run into anymore on the more modern operating systems.

It's really hard to say what's happening without the installation logs.

https://helpx.adobe.com/flash-player/kb/installation-problems-flash-player-windows.html

That said, to at least kind-of answer your question:

I don't think we have anything in the test lab that's older than an Intel Pentium Core/Celeron at this point, which covers everything shipped in the last decade.  I'd have to actually go test it to give you a definitive answer.  We're not explicitly trying to target a specific CPU, but we also are generally pretty conservative with regard to adopting new features, and will at least degrade gracefully when they're not available.

My guess is that you can probably get away with a P4/5, but I'm less confident about P3.  I can't back any of that up with actual data.  Folks in the community might be able to confirm.

At this point, I'd strongly recommend treating that machine like a kiosk.  Disconnect it from the network and just pick a set of software that works.  Having a WinXP machine attached to the Internet at this point is deeply problematic, particularly if you're doing anything like banking or healthcare-related stuff with it.

If you've got software that you like and things that you like to do with it that don't require an Internet connection, just air-gap it and keep using it.  It doesn't matter if you update it, as long as it stays off the Internet (and maybe be picky about USB memory sticks). 

In the US, you can buy a decent Chromebook for surfing the web and checking email for $200-250 (or a Win10 laptop for $400).  You might also think about just moving to a tablet or mobile device for surfing the web and checking email.  Amazon and Walmart both sell refurbished Apple hardware, so you can pick up an older iPad on the cheap.  Amazon has their own Kindle Fire tables that tend to be pretty nice hardware, and are really cheap.

Anyway, in terms of keeping that WinXP box alive, you can find archived versions of Flash Player on our Archives page.  The oldest thing we have posted is from 2010.  I'm not sure if that's going to go back far enough for you, but I'm curious to know how it works out.

[Robert Mc Dowell]

Personally I have a duo core2 Dell laptop running XP 32bit, 7 32bit, and 10 64bits.

For sure I'm surfing behind a strong firewall without any problem. I can say that on this type of machine XP stays the fastest , and still using it for my dev...

Jeromie, is the Flash 30 installer upgraded the crypto?

CBA2 J you can try to completely uninstall Flash, remove all related files reboot and reinstall....

[CBA2]

Thanks!  Lots of good advise here.  Agree, not worth too much troubleshooting, especially as Flash 29.0.0.171 works.  I have uninstalled this version and removed all old file traces and reinstalled clean w/o any issues at all.

To me it seems that it's an installer issue and not a problem with Flash 30.0.0.113 itself.  Perhaps a programmer compiled the package with a compiler requiring SSE2.  I'm saying this as I have had similar installer issues with other software providers, like Macrium.  It was resolved once they recompiled the "offending" module(s).  Just a thought.

Again, thanks for your advise which I'll study carefully.

PS: I'm posting this with one of my Win-7 computers (I have both 32 and 64-bit).  The old box runs XP Pro SP3 and I keep it for nostalgic reasons (silly, I know...) but also to run a couple of very old PC-DOS database programs that require LPT (parallel port) to print.

[Robert Mc Dowell]

There is still a lot of countries where XP is used, especially in poor countries where people don't have a clue with computer and just use it as it is, as long as it works. Microsoft must for sure think about their update that takes hours blocking your computer sometimes which is not really an evolution in 2018. On Linux for example the system is upgrading whlie you can work without any slow down.

Did you try Linux? there is everything for you, old stuff (ms dos, windows emulators etc), while the system itself is rock solid, even on P3

[jeromiec83223024 · Accepted solution]

WinXP isn't viable as a platform for using the Internet at this point, at least from a security perspective.  I liked WinXP a lot, but Win7 x64 is really the minimum bar for browsing safely at this point (32-bit address spaces make life too easy for malware authors - that holds for Linux, too).

We've made compiler upgrades over the last few months in order to pick up Microsoft's compiler-based mitigations for Spectre and Meltdown.  There's a decent chance that this is what has precipitated the change.  We wouldn't roll those changes back, and those mitigations are only available in the newer versions of the compiler.  It's hard to know if it's the compiler mitigations, or perhaps other on-by-default optimizations in the version of the compiler that we're using that might be causing the headache on the Pentium 3.

While we get that a lot of old hardware stays in use in the field, there's a limit to what we can reasonably support, and we're talking about CPUs that are well into their late teens.  Intel Core CPUs were introduced around 2008, and we absolutely support those, so we're maintaining at least a decade of hardware compatibility.  We generally take the stance with trailing-edge technology that we're not going to break it explicitly, but if it stops working at some point, we're not going to invest in keeping it alive.  In fact, I'm not even allowed to have a WinXP machine in the building at this point, much less attached to the network.

It raises an interesting point, in that we had explicit requirements for CPU hardware back when we supported mobile, but we don't appear to publish equivalent minimum system requirements for Desktop.  I don't believe that we would make a significant engineering investment to support an 17-year old CPU.

Also, it looks like global WinXP adoption is in the 2% range, down from 5% about a year ago.  It's vanishing rapidly, and the TLS/SSL technology underpinning the HTTPS protocol is down to one remaining configuration that is both considered kind of secure and actually works on WinXP.  We're definitely in the end-days of WinXP being remotely usable in the context of the larger ecosystem.

FWIW, I've got an Atari 800XL hanging around.  We used to keep it in the office running Zork.  It got a lot of love.  It's the legacy hardware in the context of the larger ecosystem where this stuff breaks down.  There's something to be said for how good the Ethernet standards have held up that you can still easily put 18-year-old machines on the network.

[Robert Mc Dowell]

True for Zork . I loved to work with Atari as a music midi sequencer attached with an AKAI sampler

[CBA2]

WinXP isn't viable as a platform for using the Internet at this point, at least from a security perspective.  I liked WinXP a lot, but Win7 x64 is really the minimum bar for browsing safely at this point ... We're definitely in the end-days of WinXP being remotely usable in the context of the larger ecosystem.

Agree, it may be time to leave my old WinXP P3 legacy box behind and catch up a bit.  But, that's another story and slightly off topic.

We've made compiler upgrades over the last few months in order to pick up Microsoft's compiler-based mitigations for Spectre and Meltdown.  There's a decent chance that this is what has precipitated the change.  We wouldn't roll those changes back...

I think that's it then.  I know, thinking is not knowing .. but, if there is a decent chance that this is causing the installer to crash, I can live with that.  I'll be happy to mark your answer as correct .. and go on with life!  Thanks a bundle, great feedback!

It's hard to know if it's the compiler mitigations, or perhaps other on-by-default optimizations in the version of the compiler that we're using that might be causing the headache on the Pentium 3.

And it's irrelevant at this time too.  It is what it is.  Again, thanks for your comments.

Postscript: I tried the Flash 30.0.0.113 .msi package and the installer advanced further than the .exe package before it stopped with an error and roll-back.  Except that it didn't roll back.  Instead, Flash 29 was totally removed and all the Flash 30 files were unpacked into C:\WINDOWS\system32\Macromed\Flash as during a normal install.  Plus the .msi install process removed all entries in the registry.

So, I exported the registry entry from another computer with working Flash 30 (not recommended) and imported it to XP's registry.  To my surprise, Flash 30 was recognized and worked normally with Firefox.  As all this was just meant to be a troubleshooting exercise, I then tried to uninstall Flash 30, but the uninstaller crashed.  Interesting experience which in my view points towards a Flash 30 installer/uninstaller package issue (and not a problem with the Flash files themselves).

Finally, after a manual removal of all Flash 30 "leftovers" I reinstalled Flash 29.0.0.171 without any problems at all.  FWIW.

[CBA2]

Re. reply #3:  you can try to completely uninstall Flash, remove all related files reboot and reinstall....

As mentioned, I tried a complete uninstall of Flash 29 (including the removal of old file/registry traces) before attempting to install Flash 30.0.0.113, however, it didn't work either.  Never passed the .exe file installer phase.

Re. reply #5:  Did you try Linux? there is everything for you, old stuff (ms dos, windows emulators etc), while the system itself is rock solid, even on P3.

No, I did not.  As I have 4 Win-based systems to maintain, partly by remote link (via TeamViewer), I don't want to go Linux.  And this is a bit off topic anyway.  Thanks for your suggestions, appreciated!