Here's why your Samsung Blu-ray player bricked itself: It downloaded an XML config file that broke the firmware Since the middle of last month, thousands of Samsung customers found their older internet-connected Blu-ray players had stopped working. In the days that followed, complaints about devices caught in an endless startup boot loop began to appear on various internet discussion boards, and videos documenting the device failure …
I feel somewhat smug now for NOT connecting any of our Samsung ‘smart’ devices, TV, Blu-Ray, Freesat box, etc., to the Internet.
I wanted these devices and what they could do, but not the endless snooping that Sammy wanted.
I do feel really sad for the poor sods caught out by this. I try not to buy Sammy stuff these days.
Because of the above, but also...
I used to be a VAR for Samsung.
I specialised in combo Sammy DVD/VHS player/recorders and HDD recorders.
After I had a number (Dozens!) of them returned for repair because the customers dared to turn of the unit at the mains switch when not in use!
Repeated ‘inrush’ current spikes destroyed a part of the power supply and these needed fixing.
Cheers Sammy, you never did pay me for the extra transport costs on those units from the customer to me and back, nor the petrol to and from your dingy repair centre (then) in Brentwood - cost me a good £500 in losses.
Pissed-off customers is great for return business.
I stopped selling Sammy gear after that.
Never regretted it.
While Sammy seems to slurp more (and more sloppily) than most, it's a wider problem with the Internet of Things in general. I don't want or need Sammy or Amazon or Google or anyone else knowing any more about my life than is necessary (and my Blu-Ray player definitely doesn't qualify.
Unfortunately it's getting harder to avoid. When I replaced my washer and dryer almost every well rated option, including Sammy's was "smart" and connected. It would be marginally more convenient to be notified when a cycle was done but not at the cost of my privacy. And goddess forbid if the software borks.
We've got a Samsung washing machine and its on its 4th motherboard but as its out of guarantee that will be its last. Not sure where the problem lies - it comes up with an error every now and then and if you turn it off and on again you can finish the halted wash. Got to know the engineer quite well as he is the only one in the area and he tried his best poor lad but the error number that came up didnt exist according to Samsung and after replacing all the parts he could think of he just tried new motherboards.
EcoBubble by any chance? The fault lies in the holes drilled in the motherboard for the component legs. They're too big, repeated vibration eventually causes dodgy contacts because of this. It's an easy fix from eBay for a recon board at about £40 from memory and only takes about 30 mins to do - but you shouldn't have to.
Samsung won't admit it is a design issue though. So we'll never have a other Samsung appliance ever again.
Do you really think Sony is any better? Sony's blurays were first in the ring with unlimited telemetry, leading the way. Plasticky Sony gear breaks all the time, though in fairness their projector series VPL-HWxx has been pretty awesome with good colour straight out of the box (argh Panasonic and Benq and Viewsonic, had mid-tier versions of those brands in before I had to admit defeat and go with Sony), and reasonably quiet.
No code should ever trust any input, whether it's user input, a remote or local file, or from a driver or anything else. The only way to safely parse is to assume that one is being fed garbage data and to defensively program to deal with such junk.
Good to know that Samsung's QA department runs such poisoned input tests against their device firmware before releasing them to customers.
Also, every internet connected device should have an externally accessible emergency reset button (like most/all home routers do), so that in the event of børkage, the storage is wiped, the device resets to defaults, and it can then safely download any needed updates (well, safely assumes that upstream has fixed the invalid file by then, of course, but the point is that you can happily reset your device once they have done so, and not need it to be sent off for repair).
Or maybe they could just do without this particular pointless telemetry crap in the first place.
There could be BluRay key updates as well. Don't forget that DRM is more important than functionality for all of these devices.
I think I might have one of these players but I gave up on its Netflix capability years ago -- old, slow and what-have-you so its not hooked up to the net any more. A Roku works for me. It might be sending my life history to Roku Center every time I touch the remote but at least they know how to write software.
I am somewhat resigned to the observed reality that just because you've tested something to destruction that doesn't guarantee it won't go TITSUP due to something unexpected - hopefully in an entertaining fashion, or in a way which leads to new scientific discoveries. A user accessible method for restoring a "last known good configuration" (which could be the same as a "factory reset") seems essential for any "smart" gadget - and is a curious omission on these Blu-ray players.
> That would, presumably, require a bit of code and a bit of storage.
True. And a button. But the on-board storage on the SoC (presumably some MediaTek/ARM-32 jobbie) might be large enough to hold a factory boot config which is used when the one in flash won't boot - or if the user holds in a button while powering up the device. So the only additional hardware needed could be a button - unless of course you can use an interface key which is already present, in which case the hardware cost would be zero (ok, make that two buttons, for a proper three-finger-salute). That "only" leaves the code... and the testing... but those should be one-off costs. And the beancounters can do one; it will cost the company more to cover mail-in "repair" costs if when their devices go TITSUP - not just in terms of logistics, manhours and materials, but in brand reputation. As demonstrated by this very thread.
> Would it need extra hardware?
If I was designing it I would include a button to purposely trigger a "factory reset" or other recovery option. But most such gadgets include user interface buttons anyway (unless they've gone all "glass"*) - I'm sure the bootloader could be made to check their state instead, negating the need for an extra reset button. E.g. "hold down [stop] and [skip back] while powering on to perform a factory reset".
*) What is it with people and touchscreens? How is it possible to prefer pressing your finger against a perfectly flat glass surface with zero feedback over pressing a distinctly tactile button!? I'm sure it's only a question of time before we get Nintendo DS style laptops with a touchscreen keyboard which doubles as a social media interface that you cannot disable.
Edit: Hell, the way things are going I predict touchscreen pianos and guitars will be a thing any day now. Just remember you heard about it here first!
Edit 2: No really; people look at new cars with all glass cockpits and go "oooh, sexy, I want that one". Only to receive a Darwin award shortly after their purchase, when trying to change the fan setting while on the motorway.
Not quite: you still need to keep some physical buttons pressed in while powering up to get the phone into recovery mode, unless of course it's an Apple device in which case you're SOL. The reason for this is quite simple: the SoC won't know that you want it to go into recovery mode unless you tell it, and to do that via the touchscreen would require the device to boot up first - so if the reason you want to enter recovery mode is that it won't boot... Physical buttons by contrast are connected (more or less) straight to the SoC's GPIO pins and can therefore be read by its low-level firmware prior to booting.
Ah, I see. I'm sorry, but I couldn't understand what you meant with
> Well yes, phones do this.
Do what? From looking at pictures* of stand alone Blu-ray players, they seem to use touch interfaces almost exclusively, but these are of the capacitive touch "button" type (as opposed to a GUI on an LCD), which should be (almost) as easy to read as physical buttons. A touch button interface often uses a dedicated chip to read the capacitive sensor inputs and translating them to logic levels, so from the SoC's perspective they look just like regular buttons. Some SoCs have native capacitive inputs, eliminating the need for an external chip. It's still an inferior technology to physical buttons though, over which it only really has two advantages: bling factor and cost saving. Some might argue capacitive touch inputs are better environmentally protected and more durable than physical buttons, but then you don't know buttons as well as I do :)
*) A photo on the web is about as close as I'd like to get to one of those things.
If its a bad cert, that may not do much to help. Once you introduce encryption and DRM, things get very difficult in the long term.
Everyone should have a proper computer to run boot services from, but failing that, maybe ethernet over phone-usb link for TFTP/HTTP boot? We can do a TFTP boot from rom quite easily and we don't need fancy security for it.
Also, what happened to "three failed boots and revert to the last firmware image" that so many PC's have?
True, but there was no problem with the input. It just contained an empty list, which is perfectly legal. Just not tested.
After software I worked on ran into something like this boot loop (not released fortunately), I added the following code: Before loading a file, the app noted that it was loading the file, then loaded it, then removed the notice. If loading the file crashed, on the next app start it would find the notice “loading file”, and in that case the file wasn’t loaded. With that method, the player would have crashed once only.
They didn't fully parse the input, because they blindly made an assumption about at least part of the input, namely that there would be a list. They did not attempt to ascertain that a) a list was present, and b) that it contained valid items.
'Parsing' doesn't mean that the XML, JSON or whatever parser doesn't bail out on you. It means that at the end of the parsing process you have all of the expected bits in the right format, along with a few optional bits, all of them fully validated and confirmed.
There's a reason why input testing often involves fuzzers, which will mix up input patterns to create a wide range of inputs, some of which will pass the XML tree parser, some of which will either pass the validators after the basic parsing. Or alternately make things catch on fire.
The volume of blocked accesses for ads and tracking was eyeopening me.
But this pi-hole strategy goes only so far - DNS over HTTPS allows it to be entirely circumvented.
Had to cook up ip address discard using iptables, which is holding so far.
But a transparent proxy is probably where it is at.
Does a standalone device such as this need to phone home (and all the rest)?
I see this incident as just the start of planned borkage by makers. In this case, Samsung got found out others will be a lot sneakier. It is so simple to send a [redacted] config file and then tell the poor sods that now have useless devices that there is no fix apart from buying a new one that will probably suffer the same fate the day after the legally mandated warranty runs out.
My home is an IoT free zone and none of my electronic kit other than my computers/phone/tablet will ever be even indirectly connected to the internet.
You beat me too it. Standalone devices, even if they can have a firmware upgrade, should not be connecting to the internet to phone home. If they are connecting it should be on the positive control of the owner for a very specific purpose.
My digital cameras get firmware updates. The manual process for updating the firmware is rather convoluted. The update is downloaded using a computer. The file is transferred to the SD card at a very specific location on the card. The SD card is loaded into the camera. Then the camera is turned on and the update is process. At no time does the camera have direct internet access to get the firmware.
To which the "correct" response should be to take the disk back to the retailer and tell them "it won't play". Some will then demonstrate that it plays in their player and tell you it's your player at fault, others will replace/refund the disk and send it back to their supplier - this is the key, it adds costs up the chain and the entire chain in retail hates costs, and by association hates actors within the system that create costs.
If LOTS of people did this, then studios turning out "stuff that won't play" would find themselves shut out of major retail businesses.
Of course, if you are told that your player is faulty, you complain to the retailer you bought it from and create costs that way. Either way, you create costs for accepting the Digital RightsRestrictions Management.
That effectively gives your player a 6 year life as that's the limit (in England at least) for civil action.
I connected my telly to the internet as the missus had a free amazon prime trial and turned it off after a few hours as we couldn't find anything worth watching on it! What did amuse me was the TV schedule listing went from (say) ITV1 program program program to ITV1 pointless bloody icon program program making it even more useless than normal.
To try and avoid the device having a lifetime of 1-2 years if some software feature requires a software update to continue working.
I know we see these devices as standalone but the reality is that many of the features require regular updates to ensure "essential" security fixes and other important updates can be applied and the device continue to operate.
However consumer goods manufacturers seem to have ignored many of the lessons about updating remote devices over the last 20+ years.
Or, you know, manufacturers could just make them work out of the box. Test them properly. Give QA sufficient resources to actually test them properrly. Novel concept, I know.
I don't care if that makes a new telly cost an extra 30, 40, 50 quid. I don't buy one every month, it's an occassional purchase, and I'd rather be confident it's been properly tested and it will actually work straight away, over a cheaper one with the highly desirable feature of randomised remote borkage.
Does a standalone device such as this need to phone home (and all the rest)?
Blu-ray players are never entirely stand-alone, because Blu-ray DRM keys are changed from time to time, and the player needs to be able to get the latest keys to decode the latest titles.
Not that any of this prevents piracy, of course, because you don't have to be able to decrypt a disk to make a bit-perfect copy of it. DRM is, and always will be, a technology for annoying the consumers of licensed media, not one for preventing consumption of unlicensed media.
Any kind of forced "phone home" updates _ARE_ highly overrated... and the IOT is _NO_ exception!
It was a lesson in "why the Micros~1 way is BAD". Meanwhile, many bricked boxes later, they were ALSO caught revealing your private viewing secrets by regularly 'phoning them home'.
*BAD* Sammy. No biscuit!
This post has been deleted by its author
I actually had a sales drone say this to me "Do you want new features? We need to move fast and break things" to which I replied with cancelling their 'service' saying we'd like the features we thought we'd bought to work before we got pointless extras.
Though I guess jokes on me because SAP bought them for $8billion...
the "move fast and break things" mantra
Very much this. I'm currently trying to curb the enthusiasm of a development team who've been assigned to me to redevelop a pretty complex test, evaluation and placement application for education. After 6 weeks of sprints and scrums, we have a beautifully designed front end web form with no security and no database design (3 databases with different users and updates, synced on a regular basis - Oooh! let's put autonumbers in every table!) and not even the vaguest of discussion on how this data will fit into the other required modules or the overall data policies of the department.
I'm retiring next year, which will be a bit of a shame, as I'd love to be around when this particular shit hits our fan.
It sounded to me more like the XML parser ran fine, didn't crash, and returned the empty list as an empty list.
However the part of the program that used the data after it was parsed expected to have a non-empty list, and then it crashed.
What was probably needed was thorough end to end unit tests that covered all eventualities. Far too many people think that if the compiler is happy with their code then it doesn't have any bugs. Compilers aren't designed to be a substitute for testing. The only way you know if something works is if you test it.
As well as development testing, there should also have been an automated test system that tried out the config first in a VM and then on actual hardware before pushing it out to the distribution server. Evidently either there wasn't, or there was but nobody tried to see what would happen on a reboot.
"As well as development testing, there should also have been an automated test system that tried out the config first in a VM and then on actual hardware before pushing it out to the distribution server. Evidently either there wasn't, or there was but nobody tried to see what would happen on a reboot."
Maybe Murphy struck and it DID pass the VM and actual unit test, but somehow something changed between the test unit and the production unit without anyone's knowledge...
"Far too many people think that if the compiler is happy with their code then it doesn't have any bugs. Compilers aren't designed to be a substitute for testing. The only way you know if something works is if you test it."
That's one of thre reasns Ada never really caught on. Too much thinking required to even make the code compile. The fact that if it did compile it probably wouldn't later have had Big Trouble with empty lists (for example) wasn't of interest to the Powers That Be, either inside or outside the company.
Good job companies in the world of safety critical systems don't have this kind of problem.
What? Some of them do? Oh dear. Time to buy one of BA's now-surplus 747s, perhaps.
"Why is the XML parser not a separate sandboxed process "
doing THAT costs more. And it won't use that crappy little 3rd party library with the cool name... that the dev found pasted over on stack overflow one day, so he could play video games instead of write code... yotta yotta yotta.
This post has been deleted by its author
Wrong.
How much, if anything, the device logs and reports back should be user defined.
And not a simply "Send telemetry yes/no" option and absolutely not a "in order to do this, Samsung needs to rape your mortal soul" dialogue where any option other than accepting disables core functionality (thus meaning it's not a free and informed choice, it's basically blackmail).
It's about time manufacturers offered two simple options. Option one - we get the device for free, they can spy all they want. Or option two - we pay for the device, and the only time the thing phones home is a simple check for updates. Anything else needs to be stamped upon. With prejudice.
I disconnected my Humax PVR from t'internet because it phones home and installs software upgrades automatically without giving the user any way to stop it. The upgrades affect the user interface, so I may or may not deem them worthwhile.
A side effect is that the device can't search for programmes any longer, but I don't care about that.
And yes, my TV predates 'smartness' being invented. It has USB connectors for media but no ethernet :)
... these internet-connected Blu-ray players in question are programmed to log their activities and send copies of this information to Samsung.
Absolutely unacceptable.
You (as the owner of the unit) should be made well aware of this by the manufacturer before you purchase it and (obviously) be able to choose to have this err ... feature? turned on or off.
Smart?
Smart my ass ...
O.
Common misconception, deliberately designed to mislead the challenged of thinking (i.e. the average consumer).
"Smart device" is one of the biggest misnomers around. These things aren't "smart" for us. They're "smart" for their manufacturers, raping and pillaging our every interaction to provide them a wealth of marketable data.
More like this please. It was great to see a fairly in-depth analysis of the reason behind the borkage.
Something so superficial (and pointless to the consumer) as telemetry bricking a device is a little painful... At the same time, I don't personally put any "locked down" machine on the internet. It just isn't worth the mess that they seem to make of themselves.
Hang on. Wasn't it Samsung who shipped laptops in which if any CMOS write went splat, it irrevocably bricked the motherboard?
You'd have thought they might have learned something from the kicking they got over that. Apparently not.
Maybe someone should go to their firmware developers' office and paint "Recovery Process" on the wall in six foot letters.
Google tells me that it would require more letters, because it looks like this:
돼지의 항문 안에 머리를 찌르십시오
I had to modify the original wording a little in order to get something that, when reverse translated, wasn't comically wrong. That's probably still wrong, but hopefully right enough that the intention is clear.
Samsung punted the home entertainment operation in 2019. Even with that, they acknowledged the problem almost immediately and do fix it for free (postage included) even out of warranty. Some should take that as an example. But samsungcloudsolutions.net is going on the NXDOMAIN list forever regardless. :-(
I mostly agree with you. Yes, they are fixing it for "free". But the user will be without their kit for likely at least 2 weeks on top of the time since it happened. Whether it's in or out of warranty is moot since Samsung caused the borkage in the first place by their own deliberate, if accidental, actions. Not only should they fix it, they should be offering something tangible by way of apology. In most jurisdictions I suspect the courts would find in favour of the customer if it came to that, and Samsung know that so they are doing the minimum they can get away with.
Fixing it for free, less the time required to package it up, take it to a collection point, parking/fuel/bus fares ( arranging child care?) and return journey. And then whatever is required to retrieve the device - visit to a collection point again, or waiting in all day for a delivery driver to drop kick it over your fence or whatever.
I'm guessing it's running Linux and bdpprog segfaulted on the NULL. Any mission critical process should catch SIGSEGV and at least log an error before it dies. Yes it may be so corrupted that crashes again anyway but at least it tried. Or failing that have a parent process that catches a crash and fails nicely even if it's just a shell script but I'm guessing that was too much effort and bdpprog was being called direct by init. Of course from a geek perspective it would be nice if any failure resulted in a drop into a shell so you could plug in a USB keyboard and try and fix it.
I'd have just expected a separate ROM copy of the known good firmware it shipped with that it could be reflashed to with either a hardware reset button or a control code of some kind (e.g. hold buttons x and y while powering it on). The added cost of doing that versus the potential cost of having to replace or repair entire units makes it a no-brainer.
PCs have dual BIOSes now for precisely that reason and it's not even the manufacturer's liability if you manage to fsck your BIOS up by upgrading it.
I think we can take as read that some beancounter will have ( if it was suggested) looked at the few pence per device that this would have costed, multiplied it by the number of sales and decided that it was far too expensive to pay for without a tangible benefit. The risks associated with not doing this are far too nebulous for accountancy.
I'd hope that the firmware required to play m2ts video from a disc wouldn't reach the hundreds of megs, though I guess they need room for…whatever software components it was phoning home to "upgrade", as well. And I guess it's capable of playing a bunch of different formats of that size of disc, so you may have a point there.
"Does a standalone device such as this need to phone home (and all the rest)?"
I thought the same thing; apparently it's for the netflix support etc. Agree 100%, if I used my bluray player just to play blurays, I would NEVER let it online. Why should I?
"I'm guessing it's running Linux and bdpprog segfaulted on the NULL. Any mission critical process should catch SIGSEGV and at least log an error before it dies."
It probably does, if you solder those wires up to the on-board serial port. You don't want to have it log a log file to the flash, then you would have your flash filled up with error logs. This is pretty stupid of Samsung, but embedded systems are tricky!
"Or failing that have a parent process that catches a crash and fails nicely"
It does catch the crash, unfortunately the "nice" failure mode is to reboot the player. Given a video player software would usually NEVER crash, if it did crash playing some bluray or whatever, would it be the player code itself, or a buggy driver? If a driver has bugged out, could you restart the driver or did it leave the system in a buggy state? Given all this a reboot is usually a reasonable recovery.
"Of course from a geek perspective it would be nice if any failure resulted in a drop into a shell so you could plug in a USB keyboard and try and fix it."
Not a bad idea actually! If a USB keyboard is plugged into your newer DVD player, bluray, even into your TV (which has a USB port anyway...), why not have it pop up into some recovery console?P
"How much, if anything, the device logs and reports back should be user defined."
Should point out on my parent's Samsung TV (.... which I now hope they don't send a bad XML file to..... ), there's like a page or two of privacy settings. Partially, it's like "that's nice, I can turn everything on or off", partially it's like "how much info is it sending to how many different vendors?" (In the interest of fairness, it has regular TV, "Samsung Plus" streaming TV... which appears to be PlutoTV's channels.... and a bunch of other streaming apps, so the privacy settings are to cover all of them I think.)
I believed that I had no choice but to go buy a new player, now I will go find the label to get a replacement thanks.
Now to the meat
Can someone please put together an idiots step by step guide on how to block these phone homes and tracking via the router.
I am asking this for millions of simple people who do not know how to block this stuff.
I tried blocking yahoo adverts but it appears they have a very large round robin list and it just trips to the next after 2 or 3 seconds.
Routers should come with all traffic blocked and you should have to open a port to allow stuff out, this take everything from these companies is disgusting.
Thanks again
"Can someone please put together an idiots step by step guide on how to block these phone homes and tracking via the router."
It's almost here in this thread already.
1) Find or buy a Raspberry Pi.
2) Follow the PiHole installation documentation (two or three lines).
Jiob done. There may even be a market opportunity here for preconfigured PiHole systems.
"Routers should come with all traffic blocked and you should have to open a port to allow stuff out, this take everything from these companies is disgusting."
They'll just disguise everything as web traffic. Block them, block the web and get complaints the Internet is broken.
If you have not already found it, I think you would enjoy lawcomic.net, a lawyer using comics to explain how the law was supposed to work and what actually happens instead.
It's standard practice in firmware code, where it's possible to brick a device, to flag a new download as "on trial". Once processing the new file has been deemed successful, it's marked as being OK.
On a subsequent boot, downloads still marked as On trial are deemed faulty, and to be ignored from then on.
Are all Samsung Internet connected bluray players affected? I've got one that's still working. Can't remember if it's downloaded a firmware update recently or not though. It's a nice enough device but the Internet services are useless and abandoned by Samsung shortly after the device was purchased
Any device for which there is any risk whatsoever of a loss of functionality due to a remote software update should have a button on it with which the user can always restore it to its initial configuration. This would have helped solve the problem with the PlayStation 4, as a nice bonus, but that's an example that shows why manufacturers will resist it unless there is government legislation.
Dunno. But I share your dislike of adverts showing up on the TV. It’s a bloody liberty, that’s what it is. When you buy an expensive 4K TV you don’t expect to be advertised at by the manufacturer. Part of my Samsung setup was an innocuous request to enter my email address. Right, like I’m actually going to do that! Since I have my own custom domain I put in samsung@mydomain.com. Turns out Samsung have a rat odour detector and refused that. But the not-so-smart TV was happy enough to accept sumsang@mydomain.com. Every time the TV powers up from cold it cheerfully informs me that sumsang@mydomain.com is logged in. They’re probably thinking that a small percentage of their purchasers will sign up to their lacklustre streaming service.
In the space of a week, my Sony Blu-ray player wont use Youtube because the O/S is Windows 7.
Also my Denon DRA-N5 internet radio wont connect to radio stations as Denon have abandoned updating the radio lists.
At least the Sony still plays discs but the Denon is a brick after 7 years good use.
Having your $100 consumer electronics brick itself is one thing.
Having your $100K+++ agricultural equipment become useless is potentially a bit more threatening.
https://www.vice.com/en_us/article/xykkkd/why-american-farmers-are-hacking-their-tractors-with-ukrainian-firmware (2017 - also covered here on El Reg from time to time iirc)
Have got a Samsung smart TV, many of the apps stopped working and the LCD has brown clouds of baked-on muck inside. Samsung 24" TV, very wobbly build quality and chronically slow EPG. My PC screen is mediocre build quality. They seem to make stuff out of nasty plastics. Samsung deep freezer had a door paintwork fault, replacement was as bad and I was charged £60+ labour for not fixing the problem. Now the digital display has a fault too. Used to like their Spinpoint hard disks. Won't entertain a Samsung mobile phone, white goods or anything else again.
Sorry, but I've seen what's real in life, and between the smoke, toxic pollen, muggers, crazy drivers, and now this so-called Coronavirus (I mean, either it's like seasonal flu or it's Captain F'n Trips and we're all dead anyway), I think I'll take the "life-like" experience I get from a good TV, thank you very much.
Samsung smart telly plus pihole here. About 20 percent of requests made by stuff on my network is blocked by the pihole, and then, the majority of that is Samsung. What I really don't understand is the value attributed to the advertising / data mining. Facebook thinks I like bikes, so I get endless adverts for luminous Spandex. I can assure you the combination of 16 stone of electrical engineer/pc nerd is not a good one with it. Data mining is overrated. Try producing knowledge, not data.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
