Last Known Copy

LEGACY NOTE · PERIOD DOCUMENT, 2007

I/O device error 0x8007045D, as asked in 2007

A Windows Vista help thread from February 2007, preserved as a period document: what the error meant on the hardware of the time, what the thread concluded, and why none of it is current advice.

Published 2026-08-06 · Updated — · 1,925 words · 10 sources · Status of facts checked 2026-08-06

A post carrying the timestamp Sun, 2007-02-18 09:36 stood at this address. It reported an optical drive, identified by the model string HL-DT-ST GWA4083B and described as an OEM LG part, that read CD-RW discs but returned I/O device error 0x8007045D for CDs and DVDs after a Windows Vista installation, having worked under Windows XP.4 Microsoft’s lifecycle record gives Windows Vista a listing start date of 2007-01-253 — twenty-four days before the question.

What follows describes that artefact and the error code it names. No procedure appears on this page. The replies the thread collected are addressed to a driver stack, a controller interface and an operating system that have all been out of support for years, and reproducing them as instructions would misrepresent period troubleshooting as advice. What survives the hardware is the code, what it identifies, and the custody problem it exposed.

What the code actually says

0x8007045D is not a Vista code, an optical-drive code, or a code about discs. It is a Win32 error number carried inside an HRESULT, and it can be taken apart with arithmetic alone.

Microsoft’s open specification for Windows error codes sets out the HRESULT layout: a severity bit, which if set “indicates a failure result”; a reserved bit, a customer bit, an NTSTATUS bit and a further reserved bit; an eleven-bit facility field, described as “an indicator of the source of the error”; and a two-byte code field holding “the remainder of the error code”.2 In 0x8007045D the severity bit is set, the facility field holds 7, and the code field holds 0x045D. Facility 7 is FACILITY_WIN32, defined in that same table as the region “reserved to map undecorated error codes into HRESULTs”.2 So the last four hex digits are an ordinary Win32 error number: 0x045D is 1117.a

Error 1117 is ERROR_IO_DEVICE. Microsoft’s system error code reference, carrying the document date 2025-07-14, gives its message as “The request could not be performed because of an I/O device error.”1 The same page’s requirements table gives a minimum supported client of Windows XP for the codes it documents, so the code the 2007 question named was not introduced by the operating system it was blamed on.

Two things follow, and both are checkable in that one table. First, the message identifies a layer rather than a cause: something below the application — the device, its bus, its driver — reported that a transfer failed, and the failure is passed upward with no statement of why. Second, Windows had more specific codes available and did not use one. In the same numeric neighbourhood sit 1110 ERROR_MEDIA_CHANGED, 1112 ERROR_NO_MEDIA_IN_DRIVE, and 1127 ERROR_DISK_OPERATION_FAILED, whose message reads “While accessing the hard disk, a disk operation failed even after retries.”1 Nothing in 1117 distinguishes an optical drive from a card reader, an external disk or a cable: the code names the failure, not the device.

What the thread contains, and what it concluded

The last capture of this address that returns HTTP 200, dated 2009-03-15, shows the question with seven replies beneath it, dated between 2007-05-30 and 2008-08-31.4 Every one of them is attributed to a poster the site’s software marked as unregistered.

Their content is summarised here and is not reproduced.b The first, more than three months after the question, asked which version of Vista was involved and pointed at a boot-time driver-signing option specific to 64-bit installations. From February 2008 the replies converged on a different suggestion: a transfer-mode setting exposed per channel on the ATA/ATAPI controller in Device Manager. One reply reported that the suggestion did not work. A later one reported that it did. The last dated reply, 2008-08-31, asked whether anyone had found a fix.

Whether the person who asked ever resolved the fault could not be verified: the thread as captured on 2009-03-15 carries no further post from the original poster, and it closes without a stated outcome.4 As a support ticket it failed. As a document it is intact, and it records something more durable than a fix — a fault whose entire evidence was one error code, discovered at the moment the discs were wanted.

The dates are what disqualify the answers as advice. Windows Vista’s original release passed out of support on 2010-04-14; mainstream support for the product ended 2012-04-11 and extended support 2017-04-12, all as published on Microsoft’s lifecycle page, whose support dates are shown in the Pacific Time Zone.3 The drive named in the question is a parallel-ATA-era OEM part. A 2008 remedy addressed to that stack is a period artefact, not a current instruction, and this publication does not restate it as one.

Why a disc error sits on a site about saved links

For the years this thread was live, a recordable disc was where a copy went when it had to exist somewhere other than the machine that made it. That is what makes 0x8007045D more than a hardware complaint: for anything written to a disc and to nowhere else, an unreadable disc is not a fault to be diagnosed but the moment the copy stops existing.

The shape of that failure is exactly the shape this publication records everywhere else. A copy is made. It is never checked. Its condition is discovered by accident, at read time, at the moment it is needed, and the discovery is not a warning but the outcome. Nothing in the interval between writing and reading reports anything at all, which is why the interval can run for years and feel like custody.

The services in this subject failed the same way on a larger scale, and one of them did so while this page was still being captured. Ma.gnolia lost its database and filesystem on 2009-01-30; its founder Larry Halff announced on 2009-02-17 that user data was irretrievable.10 The arrangement that had been treated as a backup is described in Backblaze’s account of 2009-03-02 as a file sync over FireWire with no integrity checking and no versioning, never tested to see whether the backups worked.5 That announcement falls between the last two captures of this address that return HTTP 200, dated 2009-02-11 and 2009-03-15.9 Wikipedia’s article on the service, cited here as a secondary source, records recovery tools that rebuilt some user data from web caches and other feeds, limited by what those external sources held.10 Everything anyone got back came from copies held outside the service.

The same arithmetic applies to the roster of bookmarking services this domain indexed in 2007. Of the fifty names on that roster, eight still offer somewhere to save a link, one survives read-only, eleven are documented as shut down, and thirty could not be verified (checked 2026-08-06).c A person holding a collection in one of those thirty learns which category it was in on the day they go to open it, which is the disc problem with a login form in front of it.

What replaced the 2007 answer

Not better media. NIST’s guide for librarians and archivists, published in October 2003, states that under recommended storage conditions CD-R, DVD-R and DVD+R discs “should have a life expectancy of 100 to 200 years or more”, and CD-RW, DVD-RW, DVD+RW and DVD-RAM discs “25 years or more”; for pressed ROM discs it gives a range of 20 to 100 years and notes that little information is available about them.6 Those are ranges spanning an order of magnitude, conditional on storage, and they say nothing whatever about one particular disc on one particular shelf. No published life expectancy answers the only question a holder has, which is whether the copy reads today.

What answers it is fixity: a value recorded when a copy is made, and compared against later. The National Digital Stewardship Alliance’s 2014 guidance defines fixity as the property of a file or object being unchanged, synonymous with bit-level integrity, and is blunt about the part that gets skipped:

Generating and logging fixity information to support the repair of digital objects is necessary but not sufficient for ensuring long-term access to digital information. The information must be put to use, in the form of scheduled audits of the objects against the fixity information. Additionally, replacement or repair processes must be in place. Ideally these will have been tested before being needed.7

The same document lists, among the uses of fixity information, the monitoring of hardware degradation: checks against fixity information for a set of objects that begin to fail at high rates can indicate media failure, which it adds should be treated as a supplement to hardware-specific monitoring.7 That is the identical event to 0x8007045D, met on a schedule instead of on demand, with a date attached and other copies still intact. The NDSA has maintained the practice as a graded matrix since the first Levels of Digital Preservation of April 2013, with version 2.0 released in 2019 and version 2.1 in March 2026.8

Stated as a rule rather than a prescription: a copy that must survive the loss of its original requires a recorded value taken at the moment it is written, a scheduled comparison against that value, and at least one other copy to repair from. A copy that only needs to be convenient requires none of that — and is not a backup, whatever it is called. The 2007 thread is a record of the difference being learned the expensive way, by a person who had done nothing wrong except trust a medium that reports its condition only when read.

The disc-image practice this era ran on is described at Disc images as an archiving format, 2009; the same check applied to a collection of saved links is set out at Auditing a link collection for rot.

Where to find it now

The thread at this address
A Wayback CDX query for this exact path, run 2026-08-06 — http://web.archive.org/cdx/search/cdx?url=infopirate.org/i-o-device-error-0x8007045d&matchType=exact — returns twenty-three records: twenty-two with HTTP 200, from 2007-06-14 to 2009-03-15, and one 301 dated 2021-04-23.9 No capture exists from February 2007, although the post carries a 2007-02-18 timestamp; the earliest capture of the page postdates the question by nearly four months. The replies survive only inside those captures. No export of the thread could be found in any other form (checked 2026-08-06).
The community that hosted it Gone
A Drupal-based social bookmarking site with a Q&A stream ran at this domain from early 2007; captures of this page as part of a working site stop in 2009, and no later capture shows the community operating (checked 2026-08-06). This publication holds no copy of its database, no user records and no email addresses, and there is no export endpoint to request. A former participant’s only remaining avenue is the Wayback Machine’s captures of the public pages.
Windows Vista
Out of support: original release 2010-04-14, mainstream 2012-04-11, extended 2017-04-12, per Microsoft’s lifecycle page carrying the document date 2022-11-02.3 No vendor support path exists for the fault described in the thread.
Ma.gnolia Gone
Data irretrievable from 2009-01-30, announced by its founder on 2009-02-17; no export window and no migration is recorded. Recovery tools rebuilt some user data from web caches and other feeds, bounded by what those external sources held.10 ma.gnolia.com does not resolve (checked 2026-08-06).
Discs written in 2007
Nothing collects them and nothing recovers one that no longer reads. Where a checksum was recorded at the time of writing, an unreadable copy is detectable as a failed comparison and repairable from another copy; where none was, the disc is the only witness to its own condition.

References

  1. Microsoft. “System Error Codes (1000-1299) (WinError.h).” Microsoft Learn, 2025-07-14. https://learn.microsoft.com/en-us/windows/win32/debug/system-error-codes--1000-1299- — retrieved 2026-08-06. ↩
  2. Microsoft. “[MS-ERREF]: HRESULT.” Microsoft Learn, n.d. https://learn.microsoft.com/en-us/openspecs/windows_protocols/ms-erref/0642cb2f-2075-4469-918c-4441e69c548a — retrieved 2026-08-06. ↩
  3. Microsoft. “Windows Vista — Microsoft Lifecycle.” Microsoft Learn, 2022-11-02. https://learn.microsoft.com/en-us/lifecycle/products/windows-vista — retrieved 2026-08-06. ↩
  4. Internet Archive. “I/O device error 0x8007045D | Infopirate.org” (archived capture). Wayback Machine, 2009-03-15. https://web.archive.org/web/20090315095937/http://infopirate.org/i-o-device-error-0x8007045d — retrieved 2026-08-06. ↩
  5. Backblaze. “Ma.gnolia Wilts with No Backup.” Backblaze Blog, 2009-03-02. https://www.backblaze.com/blog/magnolia-wilts-with-no-backup/ — retrieved 2026-08-06. ↩
  6. Byers, Fred R. “Care and Handling of CDs and DVDs: A Guide for Librarians and Archivists (NIST Special Publication 500-252).” National Institute of Standards and Technology and Council on Library and Information Resources, 2003-10. https://www.clir.org/pubs/reports/pub121/ — retrieved 2026-08-06. ↩
  7. National Digital Stewardship Alliance. “Checking Your Digital Content: What is Fixity, and When Should I be Checking It?” NDSA, 2014. https://www.digitalpreservation.gov/documents/NDSA-Fixity-Guidance-Report-final100214.pdf — retrieved 2026-08-06. ↩
  8. National Digital Stewardship Alliance. “Levels of Digital Preservation.” NDSA, n.d. https://www.ndsa.org/publications/levels-of-digital-preservation/ — retrieved 2026-08-06. ↩
  9. Internet Archive. “Wayback CDX Server API — BETA.” internetarchive/wayback repository, n.d. https://github.com/internetarchive/wayback/blob/master/wayback-cdx-server/README.md — retrieved 2026-08-06. ↩
  10. Wikipedia contributors. “Gnolia.” Wikipedia, n.d. Cited as a secondary source. https://en.wikipedia.org/wiki/Gnolia — retrieved 2026-08-06. ↩

Notes

  1. The decomposition is repeatable on any code of the form 0x8007xxxx: the leading 8 is the severity bit set, 007 is FACILITY_WIN32, and the remaining four hex digits are a Win32 error number to be read in decimal against the WinError.h tables. 0x045D is 1117. Codes with a different facility field are not Win32 error numbers and do not decompose this way. ↩
  2. Describing what the replies concerned, without their steps, is deliberate. The thread’s value here is documentary — what was asked, what was tried, whether it closed — and a page that lifted 2008 troubleshooting into 2026 imperative form would be asserting that it still applies to hardware and drivers nobody can obtain. ↩
  3. The census figures for the 2007 roster are the counts published in this publication’s Ledger, which carries each name’s status, source and checked date. Thirty of the fifty are recorded there as unverified rather than assigned a fate; they are listed at the unverified list. ↩

From early 2007 this address served a user-submitted Windows Vista help thread inside a Drupal social bookmarking community, posted 2007-02-18 and answered in seven comments through 2008-08-31. Captures of the page returning HTTP 200 run from 2007-06-14 to 2009-03-15; after that the record holds only a 301, dated 2021-04-23. A forum thread at forum.zive.cz points at this exact path; whether that thread is still served could not be verified (checked 2026-08-06).