Date: Tue, 16 Jul 2019 19:08:54 +0200 From: Domagoj =?UTF-8?Q?Smol=C4=8Di=C4=87?= <rank1seeker@gmail.com> To: hackers@freebsd.org Subject: For a first time completed S.M.A.R.T captive test Message-ID: <20190716190854.000061b2@gmail.com>
next in thread | raw e-mail | index | archive | help
11.2-RELEASE-p9 =46rom the first time I started to use FreeBSD and upon to just recently, wit= h smartmontools, I have NEVER successfully completed captive test. No matter which HDD or smartmontools version I used, upon initiating 'Exten= ded captive' test, I would ALWAYS get error: 'Interrupted (host reset)' This implies nothing is being mounted from device, so only it's node exist = in /dev/ and nothing "chats" with it except kernel. Stopping smartd service also didn't help. Searching on the internet, I have never found anyone succeeding with it. Just a "solutions" that it should never be used?! So I started to think a little bit out of the box ... HDD has it's OWN board with it's OWN BIOS + firmware, which actually holds = S.M.A.R.T version/ability and IT executes issued test from OS, using it's o= wn firmware to actually run a test. Once HDD receives test request from OS, HDD doesn't need OS at all! So, in order to get rid of a results like: Num Test_Description Status Remaining LifeTime(ho= urs) LBA_of_first_error # 2 Extended captive Interrupted (host reset) 90% 40743 = - And suspecting OS (kernel?!) is pestering HDD during it's captive test, thu= s interrupting it, AS SOON as captive CMD is issued and hangs occurs (it is= too late when hang passes by itself!), I've pulled out SATA DATA cable and= left SATA POWER cable attached. Hang is stopped as soon as SATA DATA cable is unplugged and it's used only = to transfer test request anyway to HDD and all HDD needs from that point on= , is JUST a power and it's "piece of mind"! RESULT: -- SMART Self-test log structure revision number 1 Num Test_Description Status Remaining LifeTime(ho= urs) LBA_of_first_error # 1 Extended captive Completed without error 00% 40744 = - # 2 Extended captive Interrupted (host reset) 90% 40743 = - -- FINALLY! =3D=3D> '# 1 Extended captive Completed without error' So ..., what to conclude from this? Does kernel really must "chat" with HDD in order to keep alive it's device = node in /dev/ or is it something else? If HDD supports captive test and during it, why it simply doesn't ignore OS= /kernel (it is up to HDD's firmware code to make that decision). Is this, I'm not even sure how to name it ..., a borderline bug? Anyway, it is a little bit "impractical" to use terminal with one hand and = with other to pull out SATA data cable. Domagoj Smol=C4=8Di=C4=87
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20190716190854.000061b2>