On Fri, Nov 25, 2022 at 05:55:17PM +0800, Herbert Xu wrote: > On Fri, Nov 25, 2022 at 09:46:04AM +0000, Giovanni Cabiddu wrote: > > > > I wanted a cap on the number of retries. What if the input is a zip > > bomb [1]? > > OK, but in that case we should just have a flat limit. In which > case it also makes no sense to double when retrying (as the limit > is so small), if the first decompression fails, we should go straight > to the limit. Just double checking if I understood correctly. At the first iteration, i.e. first call to decompress, allocate a destination buffer of 2 * src_len rounded up to 4K. If this job fails allocate a destination buffer of 128K and retry. If this fails, terminate. Thanks, -- Giovanni
