On Thu, Jun 25, 2020 at 10:57:46AM +0200, Jiri Slaby wrote:
> Hi,
>
> On 24. 06. 20, 15:18, Dan Carpenter wrote:
> > [ Kees recently removed the uninitialized_var() variable so ancient
> > code has started to print warnings. GCC doesn't warn about
> > uninitialized variables these days either. It's been buggy for
> > 10 years so it's probably not an emergency. -dan ]
> >
> > Hello Jiri Slaby,
> >
> > The patch 101b81590d8d: "tty: cyclades, cache HW version" from Jun
> > 11, 2009, leads to the following static checker warning:
> >
> > drivers/tty/cyclades.c:3805 cy_pci_probe()
> > error: uninitialized symbol 'mailbox'.
> >
> > drivers/tty/cyclades.c
> > 3640 static int cy_pci_probe(struct pci_dev *pdev,
> > 3641 const struct pci_device_id *ent)
> > 3642 {
> > 3643 struct cyclades_card *card;
> > 3644 void __iomem *addr0 = NULL, *addr2 = NULL;
> > 3645 char *card_name = NULL;
> > 3646 u32 mailbox;
> > ^^^^^^^^^^^^
> >
> > 3647 unsigned int device_id, nchan = 0, card_no, i, j;
> > 3648 unsigned char plx_ver;
> > 3649 int retval, irq;
> > 3650
> > 3651 retval = pci_enable_device(pdev);
> > 3652 if (retval) {
> > 3653 dev_err(&pdev->dev, "cannot enable device\n");
> > 3654 goto err;
> > 3655 }
> > 3656
> > 3657 /* read PCI configuration area */
> > 3658 irq = pdev->irq;
> > 3659 device_id = pdev->device & ~PCI_DEVICE_ID_MASK;
>
> It looks like you and/or your analyzer missed this line ^^?
Both of us missed it. It turns out there is a bug in Smatch and also
some missing features which triggered this warning.
Thanks for taking a look at it.
regards,
dan carpenter
