mirror of
https://github.com/007revad/Synology_enable_eunit.git
synced 2026-04-25 04:45:52 +03:00
[GH-ISSUE #67] About SiI3726 and SiI3826. #70
Labels
No labels
enhancement
pull-request
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
starred/Synology_enable_eunit#70
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @ctl26481 on GitHub (Jan 27, 2026).
Original GitHub issue: https://github.com/007revad/Synology_enable_eunit/issues/67
From the GPL'd libata-pmp.c sauce it reveals the 3826 on DX513 was faked by CPLD into 3726(the 3826 in the DX513 identifies as 0x3726 if one checks dmesg).
libata-pmp.c only specify 9705(Marvell 6Gbps) and 3xxx/3726/3826 in even older kernels so I think they used 3726 in DX5/RX4/DX510/DXC/RXC.
RX4 using SiI3726 can be seen from the photo:
https://www.newegg.com/synology-rx4/p/N82E16822108026R
They mixed 3726 and 3826(spoofed) in DX513, not sure if they did a 3726 in DX213 in the past since DX213 documentation is so scarce. But the 3826 in DX213 should identify as 3726. (Naming on DXC/RXC? Should just mean DX1211/RX1211(C=12 in hex), so they used 3726 for everything which is impressive BOM control.)
What's the difference between them? According to 45drives, they stated 3826 requires “special modification” to the kernel when they have to change the backplane BOM:
https://www.45drives.com/blog/architecture/our-biggest-design-decision-direct-wired-or-backplane/
And from this gentoo forum post(bottom one):
https://forums.gentoo.org/viewtopic-p-7229934.html
He quoted this post:
https://www.linuxquestions.org/questions/linux-hardware-18/sata-libata-link-reset-problem-with-port-multiplier-pmp-sil3132-sil3726-4175445070/#post4869234
It appears 3826 also had the identical quirk that required the same workaround with 3726... Only merged in late 2013:
https://lkml.org/lkml/2013/9/7/234
But not applied in $yn's internel kernel and DX513 was released a year ago in 2012! Maybe $yn just faked 3826 to 3726 from hardware side as "once and for all" solution without having to modify libata-pmp.c from their internal kernel.
I would assume 3826 is a costdown of 3726 or SIMG resolved the heat issue from 3726(HSBGA to QFP).
@007revad commented on GitHub (Jan 28, 2026):
My DX213 has Sil3726
@ctl26481 commented on GitHub (Jan 29, 2026):
Then it was probably spoofed("faked") from CPLD side just like DX513 with 3826.
@007revad commented on GitHub (Feb 3, 2026):
@ctl26481 Do have any idea why this script makes a DX510 show up in storage manager but not the drives in the DX510? Instead they appear as external drives in "Control Panel > External Devices".
See https://github.com/007revad/Synology_enable_eunit/issues/68