Re: [PATCH] tg3: fix return value check in tg3_read_vpd()

From: David Sterba
Date: Wed Dec 29 2010 - 08:17:36 EST


Hi,

On Monday 27 December 2010 21:19:27 Dan Carpenter wrote:
> Your fix is obviously correct, but could you describe the symptoms
> in your changelog instead of leaving it blank?  In the original code,
> negative error values are ignored so we never goto out_not_found.  Can
> pci_read_vpd() return any errors we care about besides -ENODEV?  What I
> mean is, did you find this through analysing the code or did it cause a
> bug at runtime?

Although pci_read_vpd returns ENODEV directly, there is a call to vpd->read
function which may return other error values and is eg. set to
pci_vpd_pci22_read() :

drivers/pci/access.c::pci_vpd_pci22_read()
may return -EINVAL or -EINTR directly
or -ETIMEDOUT or -EINTR via pci_vpd_pci22_wait()

Yes, other negative values besides ETIMEDOUT and EINTR are ignored and after
3 attempts the pos != TG3_NVM_VPD_LEN check goes outwards. The fixed version
allows to jump out immediately. So this does not manifest itself as a
misbehaviour at runtime. It was found by code analysis.

I'll post updated patch.

dave
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/