All of lore.kernel.org
 help / color / mirror / Atom feed
From: Ian Jackson <Ian.Jackson@eu.citrix.com>
To: xen-devel@lists.xensource.com,
	"Andrew Cooper" <andrew.cooper3@citrix.com>,
	"Roger Pau Monné" <roger.pau@citrix.com>,
	"Ian Campbell" <ian.campbell@citrix.com>,
	"Ross Lagerwall" <ross.lagerwall@citrix.com>,
	"Wei Liu" <wei.liu2@citrix.com>
Subject: Re: [PATCH 3/3] libxl: datacopier: Avoid theoretical eof/POLLHUP race
Date: Thu, 2 Apr 2015 17:29:23 +0100	[thread overview]
Message-ID: <21789.28259.607993.677511@mariner.uk.xensource.com> (raw)
In-Reply-To: <1427987045-23435-3-git-send-email-ian.jackson@eu.citrix.com>

Ian Jackson writes ("[PATCH 3/3] libxl: datacopier: Avoid theoretical eof/POLLHUP race"):
> We solve the race with a poll of the reading fd, to double-check, when
> we detect eof via read.  (This is only necessary if the caller has
> specified callback_pollhup, as otherwise POLLHUP|POLLIN - and,
> presumably, POLLIN followed perhaps by POLLHUP|POLLIN, is to be
> treated as eof anyway.)

I have been unable to reproduce this race.  On Linux, the pty master
returns EGAIN from read, not EOF, when the bootloader exits.

Below is the patch I used to test this.  I think this should repro the
bug on FreeBSD when doing "xl create" of a guest which (a) uses pygrub
(b) autoboots within 15 seconds.

Roger, can you conveniently test this ?  With the hacky patch below
you should see the bug, which should in turn be fixed by the 3/3 I am
just replying to.

Thanks,
Ian.

diff --git a/tools/libxl/libxl_aoutils.c b/tools/libxl/libxl_aoutils.c
index a08f780..5ba6d73 100644
--- a/tools/libxl/libxl_aoutils.c
+++ b/tools/libxl/libxl_aoutils.c
@@ -200,6 +200,10 @@ static void datacopier_readable(libxl__egc *egc, libxl__ev_fd *ev,
     libxl__datacopier_state *dc = CONTAINER_OF(ev, *dc, toread);
     STATE_AO_GC(dc->ao);
 
+fprintf(stderr, "DC READABLE\n");
+sleep(15);
+fprintf(stderr, "DC READABLE CONTINUING %o\n", revents);
+
     if (datacopier_pollhup_handled(egc, dc, revents, 0))
         return;
 
@@ -211,6 +215,7 @@ static void datacopier_readable(libxl__egc *egc, libxl__ev_fd *ev,
     }
     assert(revents & POLLIN);
     for (;;) {
+fprintf(stderr, "DC READABLE LOOP\n");
         while (dc->used >= dc->maxsz) {
             libxl__datacopier_buf *rm = LIBXL_TAILQ_FIRST(&dc->bufs);
             dc->used -= rm->used;
@@ -230,6 +235,7 @@ static void datacopier_readable(libxl__egc *egc, libxl__ev_fd *ev,
         int r = read(ev->fd,
                      buf->buf + buf->used,
                      sizeof(buf->buf) - buf->used);
+fprintf(stderr, "DC READABLE READ r=%d %s\n",r,strerror(errno));
         if (r < 0) {
             if (errno == EINTR) continue;
             if (errno == EWOULDBLOCK) break;

  parent reply	other threads:[~2015-04-02 16:29 UTC|newest]

Thread overview: 29+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-03-16 13:29 [PATCH v3 1/6] tools/libxl: Introduce min and max macros Ross Lagerwall
2015-03-16 13:29 ` [PATCH v3 2/6] tools/libxl: Update datacopier to support sending data only Ross Lagerwall
2015-03-16 13:29 ` [PATCH v3 3/6] tools/libxl: Avoid overrunning static buffer with prefixdata Ross Lagerwall
2015-03-16 13:29 ` [PATCH v3 4/6] tools/libxl: Allow limiting amount copied by datacopier Ross Lagerwall
2015-03-18 11:12   ` Ian Campbell
2015-03-16 13:29 ` [PATCH v3 5/6] tools/libxl: Extend datacopier to support reading into a buffer Ross Lagerwall
2015-03-18 11:18   ` Ian Campbell
2015-04-01 15:53   ` Ian Jackson
2015-04-01 15:59     ` Ian Campbell
2015-03-16 13:29 ` [PATCH v3 6/6] tools/libxl: Adjust datacopiers POLLHUP handling when the fd is also readable Ross Lagerwall
2015-03-18 11:31   ` Ian Campbell
2015-03-26 15:20   ` Roger Pau Monné
2015-03-30 10:40     ` Ian Campbell
2015-04-01 10:34       ` Roger Pau Monné
2015-04-01 14:36         ` Andrew Cooper
2015-04-02 15:03           ` [PATCH 0/3] datacopier POLLHUP fixes " Ian Jackson
2015-04-02 15:04             ` [PATCH 1/3] Revert "tools/libxl: Adjust datacopiers POLLHUP handling when the fd is also readable" Ian Jackson
2015-04-02 15:08               ` Ian Campbell
2015-04-02 15:27                 ` Ian Jackson
2015-04-02 15:10               ` Andrew Cooper
2015-04-02 15:11               ` Ian Jackson
2015-04-02 15:04             ` [PATCH 2/3] libxl: Cope with pipes which signal POLLHUP|POLLIN on read eof Ian Jackson
2015-04-02 15:13               ` Ian Campbell
2015-04-02 15:04             ` [PATCH 3/3] libxl: datacopier: Avoid theoretical eof/POLLHUP race Ian Jackson
2015-04-02 15:15               ` Ian Campbell
2015-04-02 16:29               ` Ian Jackson [this message]
2015-04-07 11:14                 ` Roger Pau Monné
2015-04-07 12:26                   ` Ian Jackson
2015-04-02 15:09             ` [PATCH 0/3] datacopier POLLHUP fixes handling when the fd is also readable Ian Jackson

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=21789.28259.607993.677511@mariner.uk.xensource.com \
    --to=ian.jackson@eu.citrix.com \
    --cc=andrew.cooper3@citrix.com \
    --cc=ian.campbell@citrix.com \
    --cc=roger.pau@citrix.com \
    --cc=ross.lagerwall@citrix.com \
    --cc=wei.liu2@citrix.com \
    --cc=xen-devel@lists.xensource.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.