The New Microsoft exFAT File-System Driver Is Set To Land With Linux 5.7

[calc]

NTFS is the best choice even if you have a Mac, just buy Paragon NTFS driver, it's not like you are poor and can't afford 20 bucks after you bought a fucking Mac so why using garbage filesystems.

FAT and exFAT are equally bad, no journaling, no compression. Not good for anything more than a basic and disposable USB flashdrive.

A major negative about using NTFS is that it is extremely slow under Linux, it uses 100% CPU on an i7 to even do ~ 20MB/s.

It also has problems with sharing between systems due to ACLs getting in the way, its better to just use FAT32 or exFAT, which is why Microsoft themselves recommend using those instead of NTFS.

[oiaohm]

You clearly have no idea wtf you are talking about.

I just installed Windows 10 onto the third partition of a system a couple days ago, and it did not wipe the rest of the partitions. And modern computers (> 2011) do not have a boot sector they have an EFI system partition on a GPT drive which allows for as many boot loaders as you want.

Not that straight forwards. Some horrible broken UEFI only load the Microsoft bootloader file names.

https://wiki.debian.org/UEFI#Force_grub-efi_installation_to_the_removable_media_path

Normally got around installing grub in the removable media path instead of proper. So yes under EFI if you implementation is broken you still end up with cat fight over bootloader and windows installer is not smart enough in it installer to check if the bootloader it overwriting is it own or someone else in UEFI. It would be good if it did check this for people who have to work around broken UEFI.

You still get new and recycled parts motherboards from china that are not UEFI just pure old school bios.

2011 was the start of common hardware being UEFI stuff but the bios only stuff is still turning up new in odd ball brand stuff. We also have in oddball brand stuff items like Linuxboot or coreboot turning up as well that gets really fun if you want to run windows. So "modern computers (> 2011)" is not quite right. It should be "general modern computers" your non general custom built server and desktop its not always case that they have UEFI.

[calc]

Not that straight forwards. Some horrible broken UEFI only load the Microsoft bootloader file names.

https://wiki.debian.org/UEFI#Force_grub-efi_installation_to_the_removable_media_path

Normally got around installing grub in the removable media path instead of proper. So yes under EFI if you implementation is broken you still end up with cat fight over bootloader and windows installer is not smart enough in it installer to check if the bootloader it overwriting is it own or someone else in UEFI. It would be good if it did check this for people who have to work around broken UEFI.

You still get new and recycled parts motherboards from china that are not UEFI just pure old school bios.

2011 was the start of common hardware being UEFI stuff but the bios only stuff is still turning up new in odd ball brand stuff. We also have in oddball brand stuff items like Linuxboot or coreboot turning up as well that gets really fun if you want to run windows. So "modern computers (> 2011)" is not quite right. It should be "general modern computers" your non general custom built server and desktop its not always case that they have UEFI.

I have seen buggy UEFI hardware that doesn't allow you to select which added entries to boot via the UEFI interface, but they worked fine after using efibootmgr to select it instead. I have never seen regular PC UEFI hardware that was hardwired to only allow booting Windows though. But if you intentionally buy 🤬 hardware you get what you want. Also that low quality of hardware generally has such broken ACPI you can't even boot Linux in the first place.

Are you sure there is PC hardware newer than 2011 that a company actually wrote an old style BIOS for instead of just using a CSM layer on top of it? I really doubt AMI/AWARD/etc wasted the effort and wrote a full old style BIOS for newer platforms that only obscure companies bother to use. But yea if you get really ancient hardware from China it would still have a BIOS.

[Alex Zu]

This seems like baseless FUD to me.

Microsoft has stated that they want exFat in the kernel, and have worked with the Linux Foundation on this. Are you telling us that they have tricked everyone except you into believing that it is ok to integrate an exFat driver?

Just look at Microsoft's exFat licensing page:

A statement «it’s OK to integrate an exFAT driver» is totally false as of today. The exFAT modules must be included into the ION definition of Linux Kernel. They’re not yet in and this is yet to happen in future.

Also retrofit of the code to earlier version of Linux is not covered by OIN use. Thus, the code still need to be accepted into the Linux Kernel, the definition of Linux shall be changed by OIN, the person/company using the exFAT implementation must be an OIN member. Once any of the 3 is false - the use will infringe the patents.

Private users can use exFAT without a big risk of being chased by Microsoft. https://lore.kernel.org/linux-fsdeve...ring-police/T/ - this exFAT implementation is out for quite a while. Big fish, i.e. OEM are under a big risk if any of 3 conditions are not met.

Also the fact, that retrofit is not legal, makes it hardly possible for big guys to use 5.7. Most of commercial hardware uses industrial grade Linux that is still based on version 3 of Linux Kernel. It may take a while for industrial grade Linux to adopt 5.7 code base.

There are a lot more aspect including legal and engineering in regard to use of exFAT in commercial devices - https://www.paragon-software.com/exfat-license/ - here’s is a link to Q&A that covers the topic.

As for private use - exFAT for Linux is around for quite a while.

* - I work for Paragon and of course my opinion is not independent. But facts that I quote - are facts.

[starshipeleven]

A major negative about using NTFS is that it is extremely slow under Linux, it uses 100% CPU on an i7 to even do ~ 20MB/s.

Your i7 has issues or you are using an ancient LTS distro, my i7-(something)-U laptop manages with 50-60 MB/s of sequential writes on an external USB hard drive with NTFS.
Also I've seen that using NTFS compression tanks performance much harder than it does on Windows.

It also has problems with sharing between systems due to ACLs getting in the way

What ACLs, both Linux and Mac NTFS drivers just don't care and remap them as "owned by user".

its better to just use FAT32 or exFAT,

filesystem integrity of NTFS is much much better than FAT32 or exFAT.

[starshipeleven]

I have never seen regular PC UEFI hardware that was hardwired to only allow booting Windows though.

A line of HP Envy laptops was kind of like that, apart from having stupid broken ACPI tables for Linux it would auto-boot Windows's efi bootloader and the only way to get it to boot anything else was to physically rename the bootloader file and put rEFInd as default bootloader in the standard place (/boot/efi/bootx64.efi)

But yeah it's not common at all.

[chithanh]

When someone wants to transfer files between Mac, Windoze and Linux systems, its the ONLY file system that all three of them fully support (read/write) AND has support for large file storage (FAT32 has a 4GB file size limit, exFAT has no such limit).

That is not true, UDF 2.01 has full read/write support in all three (+BSD+Solaris).

Many have stated the UDF file system is your universal or swiss army knife file system of choice for a portability among many operating systems, due to it's universal open source nature.

UDF implementation is not just available open source, it is an open standard (ISO 13346), unlike some others where the code is the spec...

https://git.kernel.org/pub/scm/linux...df/udf_sb.h#n9

UDF driver on Linux needs work.

https://en.wikipedia.org/wiki/Universal_Disk_Format

Remember 2.60 UDF specification was released 2005 so 15 years latter still not all current operating systems support it. UDF should be a good choice but is purely let down by implementation.

So? You can use UDF 2.01 if you want interoperability. What does UDF 2.60 give you that exFAT/NTFS has but UDF 2.01 not?

UDF is not good for cross-platform iirc. It might seem like that at first, but then you find out how Windows, macOS and Linux each differ slightly with their implementations/drivers, leads to compatibility issues.

What exactly do you mean? There are indeed permissions issues leading to inaccessible files but these are due to how the operating systems handle permissions/ACLs differently, not due to shortcoming of the filesystem driver. exFAT/NTFS drivers etc. just avoid this on Linux by disregarding permissions altogether. Which you can emulate with mode=... and umask=... mount options.

You clearly have no idea wtf you are talking about.

I just installed Windows 10 onto the third partition of a system a couple days ago, and it did not wipe the rest of the partitions.

You clearly had no idea that Windows routinely deletes partitions during install/upgrade. And that often goes wrong.

Windows 10 1607 (Anniversary Update) was widely reported to delete Linux partitions. More recently, if you upgrade to Windows 1903 from an earlier version, it will repartition the disk and remove one of the system partitions that were created during an install. And that also goes wrong sometimes.

Also retrofit of the code to earlier version of Linux is not covered by OIN use. Thus, the code still need to be accepted into the Linux Kernel, the definition of Linux shall be changed by OIN, the person/company using the exFAT implementation must be an OIN member. Once any of the 3 is false - the use will infringe the patents.

While this is technically true, after Microsoft has publicly stated that they approve exFAT going into Linux kernel royalty-free, then backpedaling and suing (corporate) users would cause a major backlash.

[Alex Zu]

after Microsoft has publicly stated that they approve exFAT going into Linux kernel royalty-free,

Their wording is extremely careful and never ever has Microsoft said exactly what you’re saying.

Again - there are lots of conditions to use it «royalty free». One which is a big issue for many - is GPL v2 (not v3). Every commercial implementation of exFAT requires code tuning for optimal performance, memory/CPU use, footprint and so on. GPL v2 requires this changes to be published. Not many commercial players will ever want to do that.

Original quote is here - https://cloudblogs.microsoft.com/ope...-linux-kernel/ - it says «where, once accepted, the code will benefit from the defensive patent commitments of OIN’s 3040+ members and licensees» - where did you find «royalty-free» ?

When it comes to private and non-commercial use - patent has never been an issue. So - what exactly has changed after this announcement ? Read carefully

[calc]

You clearly had no idea that Windows routinely deletes partitions during install/upgrade. And that often goes wrong.

Windows 10 1607 (Anniversary Update) was widely reported to delete Linux partitions. More recently, if you upgrade to Windows 1903 from an earlier version, it will repartition the disk and remove one of the system partitions that were created during an install. And that also goes wrong sometimes.

citation needed.

I've seen lots of newbies claim Windows 10 deleted partitions but have never seen it happen, and have never seen non-newbies claim it happened to them, so its almost certainly pebkac. However if you have no idea what you are doing its easy to select the wrong partition to install Windows into, or by playing in drive management once installed and formatting a Linux partition by mistake.

Also Windows 10 only uses 3 partitions to begin with and this hasn't changed over time (same since RTM in 2015): Microsoft basic area (OS area), Windows recovery environment (recovery area), and Microsoft reserved (to allow for extra write area when used on GPT). Note this doesn't include the EFI System partition required for GPT drives in general.

So what other partition was 1903 supposedly removing in the first place?

[chithanh]

Original quote is here - https://cloudblogs.microsoft.com/ope...-linux-kernel/ - it says «where, once accepted, the code will benefit from the defensive patent commitments of OIN’s 3040+ members and licensees» - where did you find «royalty-free» ?

When it comes to private and non-commercial use - patent has never been an issue. So - what exactly has changed after this announcement ? Read carefully

It is implicit in the announcement to eventually include the patents in OIN. You can read this directly on the OIN homepage:

Over 2,000 organizations have joined OIN. In support of patent non-aggression. For free access to OIN’s patents. And to gain royalty free access to Linux System patents of other OIN participants.

So if Microsoft backpedals on that by suing (commercial) exFAT users, that would go against their public announcements in a major way.

citation needed.

Also Windows 10 only uses 3 partitions to begin with and this hasn't changed over time (same since RTM in 2015): Microsoft basic area (OS area), Windows recovery environment (recovery area), and Microsoft reserved (to allow for extra write area when used on GPT).

I have witnessed it myself and you can try it yourself. Install Windows 10 RTM through 1809 on an empty disk, it will create 4 partitions, in Windows Setup they are called (in that order) "Recovery", "System", "MSR (Reserved)", and "Primary". If the "Recovery" partition is 500 MB or less (I think these versions defaulted to 450 MB) then it will be removed during upgrade, and a new 870 MB "Recovery" partition, moved to the tail of the "Primary" partition, will be created.

Note that this was the recommended layout for OEMs since Windows 10 1607:

Important: As of Windows 10, version 1607, the Recovery partition must be the next partition after the Windows partition. This ensures that winre.wim can be kept up-to-date during life of the device.

https://web.archive.org/web/20180207...sktop-editions and with 1903 they finally got their own Windows Setup to follow that rule.

The problem with removing Linux partitions is not new and certainly not PEBKAC, given how users lost their partitions without touching a partitioning tool themselves during Windows 10 upgrade, not Windows installation! And there are reports from users who never had Linux installed and still lost one partition during upgrade. It is not entirely proven that this is related to the recovery partition shenanigans, I give you that.