Re: mem leak in mii ? (fwd)

From: Bjoern A. Zeeb (bzeeb-lists_at_lists.zabbadoz.net)
Date: 12/20/04

  • Next message: Christian Brueffer: "Re: Marvell 88E8001 on sk0 and RELENG_5_3 - big problems"
    Date: Mon, 20 Dec 2004 12:23:39 +0000 (UTC)
    To: FreeBSD current mailing list <current@freebsd.org>
    
    

    Hi,

    haven't had any feedback on this....

    Can someone please review?
    Also answers to the questions would be welcome.

    Thanks.

    ---------- Forwarded message ----------
    Date: Tue, 23 Nov 2004 19:31:10 +0000 (UTC)
    From: Bjoern A. Zeeb <bzeeb-lists@lists.zabbadoz.net>
    To: John Baldwin <jhb@FreeBSD.org>
    Cc: Bjoern A. Zeeb <bzeeb-lists@lists.zabbadoz.net>,
         freebsd-current@FreeBSD.org
    Subject: Re: mem leak in mii ?

    On Mon, 22 Nov 2004, John Baldwin wrote:

    Hi,

    hope you won't get it twice; the first one didn't seem to go out...

    > On Friday 19 November 2004 06:49 pm, Bjoern A. Zeeb wrote:
    > > Hi,
    > >
    > > in sys/dev/mii/mii.c there are two calls to malloc for ivars;
    > > see for example mii_phy_probe:
    ..
    > > Where is the free for this malloc ? I cannot find it.
    > >
    > > analogous: miibus_probe ?
    >
    > It's a leak. It should be free'd when the miibus device is destroyed. Here's
    > a possible fix:

    could you please review this one ? Should plug both of the memleaks;
    also for more error cases.

    notes:

    * mii doesn't ssem to be very error corrective and reporting; as
      others currently also seem to be debugging problems with
      undetectable PHYs I added some error handling in those places that
      I touched anyway.
    * in miibus_probe in the loop there is the possibility - and the comment
      above the functions also talks about this - that we find more than
      one PHY ? I currrently doubt that but I don't know for sure.
      As device_add_child may return NULL we cannot check for that; I had
      seen some inconsistency while debugging the BMSR_MEDIAMASK check so
      I added the count variable for this to have a reliable state.
    * all PHY drivers currently seem to use mii_phy_detach for
      device_detach. If any implements his own function it will be
      responsible for freeing the ivars allocated in miibus_probe. This
      should perhaps be documented somewhere ?

    patch can also be found at
    http://sources.zabbadoz.net/freebsd/patchset/mii-memleaks.diff

    Index: mii.c
    ===================================================================
    RCS file: /local/mirror/FreeBSD/r/ncvs/src/sys/dev/mii/mii.c,v
    retrieving revision 1.20
    diff -u -p -r1.20 mii.c
    --- mii.c 15 Aug 2004 06:24:40 -0000 1.20
    +++ mii.c 23 Nov 2004 17:08:58 -0000
    @@ -111,7 +111,7 @@ miibus_probe(dev)
             struct mii_attach_args ma, *args;
             struct mii_data *mii;
             device_t child = NULL, parent;
    - int bmsr, capmask = 0xFFFFFFFF;
    + int count = 0, bmsr, capmask = 0xFFFFFFFF;

             mii = device_get_softc(dev);
             parent = device_get_parent(dev);
    @@ -145,12 +145,26 @@ miibus_probe(dev)

                     args = malloc(sizeof(struct mii_attach_args),
                         M_DEVBUF, M_NOWAIT);
    + if (args == NULL) {
    + device_printf(dev, "%s: memory allocation failure, "
    + "phyno %d", __func__, ma.mii_phyno);
    + continue;
    + }
                     bcopy((char *)&ma, (char *)args, sizeof(ma));
                     child = device_add_child(dev, NULL, -1);
    + if (child == NULL) {
    + free(args, M_DEVBUF);
    + device_printf(dev, "%s: device_add_child failed",
    + __func__);
    + continue;
    + }
                     device_set_ivars(child, args);
    + count++;
    + /* XXX should we break here or is it really possible
    + * to find more then one PHY ? */
             }

    - if (child == NULL)
    + if (count == 0)
                     return(ENXIO);

             device_set_desc(dev, "MII bus");
    @@ -173,12 +187,15 @@ miibus_attach(dev)
              */
             mii->mii_ifp = device_get_softc(device_get_parent(dev));
             v = device_get_ivars(dev);
    + if (v == NULL)
    + return (ENXIO);
             ifmedia_upd = v[0];
             ifmedia_sts = v[1];
    + device_set_ivars(dev, NULL);
    + free(v, M_DEVBUF);
             ifmedia_init(&mii->mii_media, IFM_IMASK, ifmedia_upd, ifmedia_sts);
    - bus_generic_attach(dev);

    - return(0);
    + return (bus_generic_attach(dev));
     }

     int
    @@ -186,8 +203,14 @@ miibus_detach(dev)
             device_t dev;
     {
             struct mii_data *mii;
    + void *v;

             bus_generic_detach(dev);
    + v = device_get_ivars(dev);
    + if (v != NULL) {
    + device_set_ivars(dev, NULL);
    + free(v, M_DEVBUF);
    + }
             mii = device_get_softc(dev);
             ifmedia_removeall(&mii->mii_media);
             mii->mii_ifp = NULL;
    @@ -305,12 +328,15 @@ mii_phy_probe(dev, child, ifmedia_upd, i
             int bmsr, i;

             v = malloc(sizeof(vm_offset_t) * 2, M_DEVBUF, M_NOWAIT);
    - if (v == 0) {
    + if (v == NULL)
                     return (ENOMEM);
    - }
             v[0] = ifmedia_upd;
             v[1] = ifmedia_sts;
             *child = device_add_child(dev, "miibus", -1);
    + if (*child == NULL) {
    + free(v, M_DEVBUF);
    + return (ENXIO);
    + }
             device_set_ivars(*child, v);

             for (i = 0; i < MII_NPHY; i++) {
    @@ -324,14 +350,22 @@ mii_phy_probe(dev, child, ifmedia_upd, i
             }

             if (i == MII_NPHY) {
    + device_set_ivars(dev, NULL);
    + free(v, M_DEVBUF);
                     device_delete_child(dev, *child);
                     *child = NULL;
                     return(ENXIO);
             }

    - bus_generic_attach(dev);
    + i = bus_generic_attach(dev);

    - return(0);
    + v = device_get_ivars(*child);
    + if (v != NULL) {
    + device_set_ivars(*child, NULL);
    + free(v, M_DEVBUF);
    + }
    +
    + return (i);
     }

     /*
    Index: mii_physubr.c
    ===================================================================
    RCS file: /local/mirror/FreeBSD/r/ncvs/src/sys/dev/mii/mii_physubr.c,v
    retrieving revision 1.21
    diff -u -p -r1.21 mii_physubr.c
    --- mii_physubr.c 29 May 2004 18:09:10 -0000 1.21
    +++ mii_physubr.c 23 Nov 2004 17:07:30 -0000
    @@ -522,7 +522,13 @@ int
     mii_phy_detach(device_t dev)
     {
             struct mii_softc *sc;
    + void *args;

    + args = device_get_ivars(dev);
    + if (args != NULL) {
    + device_set_ivars(dev, NULL);
    + free(args, M_DEVBUF);
    + }
             sc = device_get_softc(dev);
             mii_phy_down(sc);
             sc->mii_dev = NULL;

    -- 
    Bjoern A. Zeeb				bzeeb at Zabbadoz dot NeT
    _______________________________________________
    freebsd-current@freebsd.org mailing list
    http://lists.freebsd.org/mailman/listinfo/freebsd-current
    To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"
    _______________________________________________
    freebsd-net@freebsd.org mailing list
    http://lists.freebsd.org/mailman/listinfo/freebsd-net
    To unsubscribe, send any mail to "freebsd-net-unsubscribe@freebsd.org"
    

  • Next message: Christian Brueffer: "Re: Marvell 88E8001 on sk0 and RELENG_5_3 - big problems"

    Relevant Pages

    • Re: mem leak in mii ? (fwd)
      ... Can someone please review? ... Subject: mem leak in mii? ... one PHY? ...
      (freebsd-current)
    • Re: mem leak in mii ? (fwd)
      ... third and last call for review and comments. ... mii doesn't ssem to be very error corrective and reporting; as others currently also seem to be debugging problems with undetectable PHYs I added some error handling in those places that I touched anyway. ... args = malloc, ... device_delete_child(dev, ...
      (freebsd-current)
    • Re: mem leak in mii ? (fwd)
      ... third and last call for review and comments. ... mii doesn't ssem to be very error corrective and reporting; as others currently also seem to be debugging problems with undetectable PHYs I added some error handling in those places that I touched anyway. ... args = malloc, ... device_delete_child(dev, ...
      (freebsd-net)