git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [PATCH 01/10] strbuf: Add strbuf_read_noblock

From
Jeff King <peff@peff.net>
Date
Sep 17, 2015, 16:30 UTC
Message-ID
<20150917163012.GB25837@sigill.intra.peff.net>
In-Reply-To
<xmqqtwqtja6j.fsf@gitster.mtv.corp.google.com>
On Thu, Sep 17, 2015 at 09:13:40AM -0700, Junio C Hamano wrote:
Show 10 quoted lines
> And your new caller that does O_NONBLOCK wants to do more than
> looping upon EWOULDBLOCK.  It certainly would not want us to loop
> here.
> 
> So I wonder if you can just O_NONBLOCK the fd and use the usual
> strbuf_read(), i.e. without any change in this patch, and update
> xread() to _unconditionally_ return when read(2) says EAGAIN or
> EWOULDBLOCK.
> 
> What would that break?

Certainly anybody who does not realize their descriptor is O_NONBLOCK and is using the spinning for correctness. I tend to think that such sites are wrong, though, and would benefit from us realizing they are spinning.

But I think you can't quite get away with leaving strbuf_read untouched in this case. On error, it wants to restore the original value of the strbuf before the strbuf_read call. Which means that we throw away anything read into the strbuf before we get EAGAIN, and the caller never gets to see it.

So I think we would probably want to treat EAGAIN specially: return -1 to signal to the caller but _don't_ truncate the strbuf.

Arguably we should actually return the number of bytes we _did_ read, but then caller cannot easily tell the difference between EOF and EAGAIN.

-Peff
Previous: Junio C HamanoNext: Junio C Hamano
Message 4 of 37 in “fetch submodules in parallel and a preview on parallel "submodule update"”
  1. 00/10 fetch submodules in parallel and a preview on parallel "submodule update"Stefan Beller, Sep 17, 2015
  2. 01/10 strbuf: Add strbuf_read_noblockStefan Beller, Sep 17, 2015
  3. Junio C HamanoSep 17, 2015
  4. Jeff KingSep 17, 2015
  5. Junio C HamanoSep 17, 2015
  6. Stefan BellerSep 17, 2015
  7. Jeff KingSep 17, 2015
  8. Junio C HamanoSep 17, 2015
  9. Jeff KingSep 17, 2015
  10. Stefan BellerSep 17, 2015
  11. Jeff KingSep 17, 2015
  12. Stefan BellerSep 17, 2015
  13. Jeff KingSep 17, 2015
  14. Stefan BellerSep 17, 2015
  15. Junio C HamanoSep 17, 2015
  16. Junio C HamanoSep 17, 2015
  17. Jeff KingSep 17, 2015
  18. Stefan BellerSep 17, 2015
  19. 02/10 run-command: factor out return value computationStefan Beller, Sep 17, 2015
  20. Jeff KingSep 17, 2015
  21. 03/10 run-command: add an asynchronous parallel child processorStefan Beller, Sep 17, 2015
  22. Junio C HamanoSep 17, 2015
  23. Stefan BellerSep 17, 2015
  24. Junio C HamanoSep 18, 2015
  25. Stefan BellerSep 18, 2015
  26. 04/10 fetch_populated_submodules: use new parallel job processingStefan Beller, Sep 17, 2015
  27. 05/10 submodules: Allow parallel fetching, add tests and documentationStefan Beller, Sep 17, 2015
  28. 06/10 git submodule update: Redirect any output to stderrStefan Beller, Sep 17, 2015
  29. Eric SunshineSep 17, 2015
  30. Stefan BellerSep 17, 2015
  31. 07/10 git submodule update: pass --prefix only with a non empty prefixStefan Beller, Sep 17, 2015
  32. Eric SunshineSep 17, 2015
  33. 08/10 git submodule update: cmd_update_recursiveStefan Beller, Sep 17, 2015
  34. 09/10 git submodule update: cmd_update_recursiveStefan Beller, Sep 17, 2015
  35. Eric SunshineSep 17, 2015
  36. 10/10 git submodule update: cmd_update_fetchStefan Beller, Sep 17, 2015
  37. Jacob KellerSep 17, 2015

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.